Never stop coding. Free MIT AI gateway: one endpoint, 375 providers (150+ free), 1,400+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 900+ contributors
TypeScript a2aai-agentsai-gatewayanthropicclaudeclaude-codeclinecodexcopilotcursordeepseekfree-aigeminikimillm-gatewaymcpopenaiopenai-proxyqwentoken-saver
feat(github): add targetFormat openai-responses to all GitHub models (#2122)
Abhinav Kumara3e9934fa05f
diff --git a/docs/i18n/ar/CHANGELOG.md b/docs/i18n/ar/CHANGELOG.mdindex 197fafc273..7b98a2c891 100644--- a/docs/i18n/ar/CHANGELOG.md+++ b/docs/i18n/ar/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### الأمان+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### التوثيق+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### التوثيق+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### الأمان+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### الأمان+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### التوثيق+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### الأمان+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### التوثيق+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### الأمان+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### الميزات+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### الأمان+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### الميزات+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### الميزات+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### الأمان+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### الأمان+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### التوثيق+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### التوثيق+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### الميزات+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### الميزات+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### الميزات+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### الأمان+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### الميزات+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### الميزات+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### الميزات+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### الميزات+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### الميزات+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### الميزات+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### الميزات+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### الميزات+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### الأمان+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### الميزات+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### الميزات+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### الميزات+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### الميزات+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### الميزات+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### الميزات+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### التوثيق+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### الميزات+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/bg/CHANGELOG.md b/docs/i18n/bg/CHANGELOG.mdindex 78b3ebb64a..8d16275353 100644--- a/docs/i18n/bg/CHANGELOG.md+++ b/docs/i18n/bg/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Сигурност+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Документация+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Документация+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Сигурност+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Сигурност+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Документация+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Сигурност+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Документация+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Сигурност+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Функции+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Сигурност+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Функции+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Функции+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Сигурност+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Сигурност+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Документация+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Документация+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Функции+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Функции+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Функции+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Сигурност+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Функции+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Функции+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Функции+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Функции+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Функции+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Функции+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Функции+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Функции+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Сигурност+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Функции+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Функции+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Функции+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Функции+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Функции+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Функции+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Документация+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Функции+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/bn/CHANGELOG.md b/docs/i18n/bn/CHANGELOG.mdindex 4342edd2a8..70ec0dfa52 100644--- a/docs/i18n/bn/CHANGELOG.md+++ b/docs/i18n/bn/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/cs/CHANGELOG.md b/docs/i18n/cs/CHANGELOG.mdindex a1d8696409..ba38f13de3 100644--- a/docs/i18n/cs/CHANGELOG.md+++ b/docs/i18n/cs/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Bezpečnost+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentace+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentace+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Bezpečnost+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Bezpečnost+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentace+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Bezpečnost+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentace+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Bezpečnost+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funkce+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Bezpečnost+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funkce+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funkce+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Bezpečnost+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Bezpečnost+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentace+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentace+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funkce+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funkce+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funkce+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Bezpečnost+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funkce+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funkce+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funkce+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funkce+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funkce+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funkce+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funkce+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funkce+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Bezpečnost+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funkce+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funkce+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funkce+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funkce+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funkce+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funkce+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentace+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funkce+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/da/CHANGELOG.md b/docs/i18n/da/CHANGELOG.mdindex ac422e7b31..113ee1387d 100644--- a/docs/i18n/da/CHANGELOG.md+++ b/docs/i18n/da/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Sikkerhed+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentation+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentation+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Sikkerhed+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Sikkerhed+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentation+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Sikkerhed+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentation+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Sikkerhed+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funktioner+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Sikkerhed+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funktioner+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funktioner+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Sikkerhed+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Sikkerhed+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentation+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentation+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funktioner+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funktioner+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funktioner+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Sikkerhed+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funktioner+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funktioner+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funktioner+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funktioner+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funktioner+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funktioner+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funktioner+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funktioner+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Sikkerhed+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funktioner+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funktioner+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funktioner+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funktioner+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funktioner+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funktioner+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentation+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funktioner+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/de/CHANGELOG.md b/docs/i18n/de/CHANGELOG.mdindex aad9040459..1f5f469dfb 100644--- a/docs/i18n/de/CHANGELOG.md+++ b/docs/i18n/de/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Sicherheit+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentation+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentation+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Sicherheit+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Sicherheit+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentation+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Sicherheit+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentation+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Sicherheit+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funktionen+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Sicherheit+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funktionen+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funktionen+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Sicherheit+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Sicherheit+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentation+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentation+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funktionen+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funktionen+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funktionen+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Sicherheit+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funktionen+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funktionen+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funktionen+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funktionen+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funktionen+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funktionen+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funktionen+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funktionen+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Sicherheit+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funktionen+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funktionen+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funktionen+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funktionen+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funktionen+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funktionen+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentation+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funktionen+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/es/CHANGELOG.md b/docs/i18n/es/CHANGELOG.mdindex f1ac4292ee..6f3c269fe3 100644--- a/docs/i18n/es/CHANGELOG.md+++ b/docs/i18n/es/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/fa/CHANGELOG.md b/docs/i18n/fa/CHANGELOG.mdindex 26340b8f4d..e81105040d 100644--- a/docs/i18n/fa/CHANGELOG.md+++ b/docs/i18n/fa/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/fi/CHANGELOG.md b/docs/i18n/fi/CHANGELOG.mdindex d094cdea0e..e64bc2060b 100644--- a/docs/i18n/fi/CHANGELOG.md+++ b/docs/i18n/fi/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Turvallisuus+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentaatio+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentaatio+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Turvallisuus+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Turvallisuus+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentaatio+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Turvallisuus+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentaatio+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Turvallisuus+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Ominaisuudet+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Turvallisuus+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Ominaisuudet+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Ominaisuudet+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Turvallisuus+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Turvallisuus+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentaatio+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentaatio+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Ominaisuudet+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Ominaisuudet+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Ominaisuudet+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Turvallisuus+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Ominaisuudet+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Ominaisuudet+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Ominaisuudet+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Ominaisuudet+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Ominaisuudet+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Ominaisuudet+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Ominaisuudet+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Ominaisuudet+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Turvallisuus+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Ominaisuudet+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Ominaisuudet+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Ominaisuudet+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Ominaisuudet+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Ominaisuudet+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Ominaisuudet+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentaatio+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Ominaisuudet+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/fr/CHANGELOG.md b/docs/i18n/fr/CHANGELOG.mdindex d6e3fa893f..f086b04edf 100644--- a/docs/i18n/fr/CHANGELOG.md+++ b/docs/i18n/fr/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Sécurité+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentation+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentation+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Sécurité+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Sécurité+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentation+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Sécurité+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Sécurité+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Fonctionnalités+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Sécurité+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Fonctionnalités+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Fonctionnalités+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Sécurité+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Sécurité+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentation+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentation+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Fonctionnalités+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Fonctionnalités+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Fonctionnalités+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Sécurité+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Fonctionnalités+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Fonctionnalités+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Fonctionnalités+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Fonctionnalités+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Fonctionnalités+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Fonctionnalités+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Fonctionnalités+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Fonctionnalités+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Sécurité+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Fonctionnalités+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Fonctionnalités+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Fonctionnalités+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Fonctionnalités+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Fonctionnalités+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Fonctionnalités+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentation+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Fonctionnalités+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/gu/CHANGELOG.md b/docs/i18n/gu/CHANGELOG.mdindex b3ad23f5a8..7529472667 100644--- a/docs/i18n/gu/CHANGELOG.md+++ b/docs/i18n/gu/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/he/CHANGELOG.md b/docs/i18n/he/CHANGELOG.mdindex 45343dce03..f727b787fa 100644--- a/docs/i18n/he/CHANGELOG.md+++ b/docs/i18n/he/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### אבטחה+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### תיעוד+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### תיעוד+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### אבטחה+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### אבטחה+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### תיעוד+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### אבטחה+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### תיעוד+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### אבטחה+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### תכונות+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### אבטחה+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### תכונות+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### תכונות+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### אבטחה+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### אבטחה+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### תיעוד+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### תיעוד+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### תכונות+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### תכונות+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### תכונות+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### אבטחה+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### תכונות+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### תכונות+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### תכונות+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### תכונות+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### תכונות+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### תכונות+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### תכונות+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### תכונות+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### אבטחה+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### תכונות+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### תכונות+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### תכונות+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### תכונות+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### תכונות+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### תכונות+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### תיעוד+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### תכונות+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/hi/CHANGELOG.md b/docs/i18n/hi/CHANGELOG.mdindex a02d0e904b..64617e3af2 100644--- a/docs/i18n/hi/CHANGELOG.md+++ b/docs/i18n/hi/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### सुरक्षा+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### दस्तावेज़+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### दस्तावेज़+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### सुरक्षा+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### सुरक्षा+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### दस्तावेज़+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### सुरक्षा+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### दस्तावेज़+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### सुरक्षा+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### विशेषताएं+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### सुरक्षा+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### विशेषताएं+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### विशेषताएं+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### सुरक्षा+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### सुरक्षा+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### दस्तावेज़+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### दस्तावेज़+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### विशेषताएं+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### विशेषताएं+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### विशेषताएं+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### सुरक्षा+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### विशेषताएं+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### विशेषताएं+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### विशेषताएं+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### विशेषताएं+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### विशेषताएं+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### विशेषताएं+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### विशेषताएं+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### विशेषताएं+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### सुरक्षा+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### विशेषताएं+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### विशेषताएं+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### विशेषताएं+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### विशेषताएं+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### विशेषताएं+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### विशेषताएं+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### दस्तावेज़+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### विशेषताएं+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/hu/CHANGELOG.md b/docs/i18n/hu/CHANGELOG.mdindex b63e8b3de2..f6e7f96e4b 100644--- a/docs/i18n/hu/CHANGELOG.md+++ b/docs/i18n/hu/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Biztonság+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentáció+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentáció+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Biztonság+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Biztonság+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentáció+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Biztonság+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentáció+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Biztonság+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funkciók+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Biztonság+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funkciók+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funkciók+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Biztonság+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Biztonság+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentáció+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentáció+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funkciók+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funkciók+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funkciók+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Biztonság+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funkciók+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funkciók+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funkciók+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funkciók+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funkciók+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funkciók+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funkciók+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funkciók+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Biztonság+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funkciók+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funkciók+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funkciók+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funkciók+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funkciók+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funkciók+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentáció+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funkciók+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/id/CHANGELOG.md b/docs/i18n/id/CHANGELOG.mdindex 45562bfcbd..678ec3ed07 100644--- a/docs/i18n/id/CHANGELOG.md+++ b/docs/i18n/id/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Keamanan+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentasi+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentasi+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Keamanan+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Keamanan+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentasi+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Keamanan+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentasi+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Keamanan+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Fitur+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Keamanan+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Fitur+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Fitur+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Keamanan+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Keamanan+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentasi+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentasi+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Fitur+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Fitur+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Fitur+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Keamanan+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Fitur+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Fitur+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Fitur+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Fitur+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Fitur+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Fitur+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Fitur+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Fitur+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Keamanan+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Fitur+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Fitur+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Fitur+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Fitur+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Fitur+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Fitur+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentasi+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Fitur+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/it/CHANGELOG.md b/docs/i18n/it/CHANGELOG.mdindex 92a5337a76..b0b64fc72d 100644--- a/docs/i18n/it/CHANGELOG.md+++ b/docs/i18n/it/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Sicurezza+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentazione+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentazione+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Sicurezza+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Sicurezza+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentazione+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Sicurezza+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentazione+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Sicurezza+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funzionalità+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Sicurezza+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funzionalità+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funzionalità+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Sicurezza+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Sicurezza+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentazione+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentazione+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funzionalità+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funzionalità+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funzionalità+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Sicurezza+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funzionalità+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funzionalità+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funzionalità+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funzionalità+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funzionalità+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funzionalità+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funzionalità+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funzionalità+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Sicurezza+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funzionalità+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funzionalità+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funzionalità+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funzionalità+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funzionalità+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funzionalità+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentazione+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funzionalità+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/ja/CHANGELOG.md b/docs/i18n/ja/CHANGELOG.mdindex 3081a2c6f8..4390f70a3d 100644--- a/docs/i18n/ja/CHANGELOG.md+++ b/docs/i18n/ja/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### セキュリティ+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### ドキュメント+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### ドキュメント+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### セキュリティ+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### セキュリティ+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### ドキュメント+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### セキュリティ+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### ドキュメント+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### セキュリティ+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### 機能+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### セキュリティ+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### 機能+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### 機能+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### セキュリティ+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### セキュリティ+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### ドキュメント+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### ドキュメント+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### 機能+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### 機能+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### 機能+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### セキュリティ+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### 機能+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### 機能+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### 機能+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### 機能+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### 機能+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### 機能+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### 機能+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### 機能+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### セキュリティ+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### 機能+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### 機能+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### 機能+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### 機能+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### 機能+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### 機能+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### ドキュメント+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### 機能+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/ko/CHANGELOG.md b/docs/i18n/ko/CHANGELOG.mdindex 834983612b..a65c0ec7b1 100644--- a/docs/i18n/ko/CHANGELOG.md+++ b/docs/i18n/ko/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### 보안+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### 문서+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### 문서+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### 보안+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### 보안+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### 문서+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### 보안+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### 문서+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### 보안+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### 기능+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### 보안+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### 기능+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### 기능+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### 보안+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### 보안+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### 문서+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### 문서+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### 기능+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### 기능+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### 기능+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### 보안+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### 기능+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### 기능+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### 기능+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### 기능+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### 기능+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### 기능+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### 기능+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### 기능+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### 보안+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### 기능+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### 기능+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### 기능+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### 기능+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### 기능+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### 기능+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### 문서+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### 기능+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/mr/CHANGELOG.md b/docs/i18n/mr/CHANGELOG.mdindex e6333cb9a3..acca4a8793 100644--- a/docs/i18n/mr/CHANGELOG.md+++ b/docs/i18n/mr/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/ms/CHANGELOG.md b/docs/i18n/ms/CHANGELOG.mdindex 26db8348e9..7c0ffc85e9 100644--- a/docs/i18n/ms/CHANGELOG.md+++ b/docs/i18n/ms/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Keselamatan+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentasi+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentasi+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Keselamatan+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Keselamatan+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentasi+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Keselamatan+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentasi+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Keselamatan+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Ciri-ciri+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Keselamatan+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Ciri-ciri+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Ciri-ciri+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Keselamatan+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Keselamatan+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentasi+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentasi+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Ciri-ciri+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Ciri-ciri+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Ciri-ciri+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Keselamatan+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Ciri-ciri+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Ciri-ciri+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Ciri-ciri+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Ciri-ciri+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Ciri-ciri+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Ciri-ciri+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Ciri-ciri+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Ciri-ciri+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Keselamatan+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Ciri-ciri+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Ciri-ciri+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Ciri-ciri+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Ciri-ciri+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Ciri-ciri+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Ciri-ciri+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentasi+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Ciri-ciri+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/nl/CHANGELOG.md b/docs/i18n/nl/CHANGELOG.mdindex b77a52ec5b..003f5fa803 100644--- a/docs/i18n/nl/CHANGELOG.md+++ b/docs/i18n/nl/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Beveiliging+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentatie+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentatie+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Beveiliging+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Beveiliging+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentatie+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Beveiliging+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentatie+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Beveiliging+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Functies+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Beveiliging+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Functies+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Functies+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Beveiliging+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Beveiliging+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentatie+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentatie+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Functies+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Functies+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Functies+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Beveiliging+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Functies+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Functies+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Functies+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Functies+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Functies+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Functies+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Functies+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Functies+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Beveiliging+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Functies+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Functies+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Functies+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Functies+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Functies+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Functies+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentatie+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Functies+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/no/CHANGELOG.md b/docs/i18n/no/CHANGELOG.mdindex 4ae7e72748..508c91fa07 100644--- a/docs/i18n/no/CHANGELOG.md+++ b/docs/i18n/no/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Sikkerhet+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentasjon+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentasjon+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Sikkerhet+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Sikkerhet+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentasjon+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Sikkerhet+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentasjon+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Sikkerhet+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funksjoner+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Sikkerhet+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funksjoner+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funksjoner+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Sikkerhet+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Sikkerhet+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentasjon+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentasjon+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funksjoner+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funksjoner+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funksjoner+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Sikkerhet+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funksjoner+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funksjoner+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funksjoner+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funksjoner+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funksjoner+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funksjoner+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funksjoner+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funksjoner+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Sikkerhet+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funksjoner+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funksjoner+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funksjoner+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funksjoner+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funksjoner+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funksjoner+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentasjon+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funksjoner+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/phi/CHANGELOG.md b/docs/i18n/phi/CHANGELOG.mdindex c61c70929e..0f9b6df913 100644--- a/docs/i18n/phi/CHANGELOG.md+++ b/docs/i18n/phi/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentasyon+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentasyon+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentasyon+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentasyon+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Mga Tampok+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Mga Tampok+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Mga Tampok+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentasyon+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentasyon+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Mga Tampok+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Mga Tampok+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Mga Tampok+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Mga Tampok+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Mga Tampok+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Mga Tampok+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Mga Tampok+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Mga Tampok+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Mga Tampok+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Mga Tampok+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Mga Tampok+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Mga Tampok+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Mga Tampok+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Mga Tampok+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Mga Tampok+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Mga Tampok+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Mga Tampok+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentasyon+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Mga Tampok+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/pl/CHANGELOG.md b/docs/i18n/pl/CHANGELOG.mdindex 07fd8e7c33..fec2b26545 100644--- a/docs/i18n/pl/CHANGELOG.md+++ b/docs/i18n/pl/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Bezpieczeństwo+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentacja+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentacja+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Bezpieczeństwo+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Bezpieczeństwo+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentacja+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Bezpieczeństwo+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentacja+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Bezpieczeństwo+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funkcje+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Bezpieczeństwo+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funkcje+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funkcje+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Bezpieczeństwo+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Bezpieczeństwo+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentacja+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentacja+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funkcje+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funkcje+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funkcje+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Bezpieczeństwo+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funkcje+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funkcje+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funkcje+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funkcje+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funkcje+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funkcje+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funkcje+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funkcje+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Bezpieczeństwo+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funkcje+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funkcje+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funkcje+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funkcje+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funkcje+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funkcje+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentacja+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funkcje+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/pt-BR/CHANGELOG.md b/docs/i18n/pt-BR/CHANGELOG.mdindex fc12e01ff5..1b7f7b26bc 100644--- a/docs/i18n/pt-BR/CHANGELOG.md+++ b/docs/i18n/pt-BR/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Segurança+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentação+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentação+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Segurança+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Segurança+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentação+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Segurança+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentação+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Segurança+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Segurança+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Segurança+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Segurança+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentação+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentação+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Segurança+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Segurança+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentação+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/pt/CHANGELOG.md b/docs/i18n/pt/CHANGELOG.mdindex 2f1c91ed28..cfd9f8dd54 100644--- a/docs/i18n/pt/CHANGELOG.md+++ b/docs/i18n/pt/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Segurança+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentação+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentação+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Segurança+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Segurança+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentação+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Segurança+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentação+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Segurança+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Segurança+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Segurança+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Segurança+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentação+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentação+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Segurança+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Segurança+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentação+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/ro/CHANGELOG.md b/docs/i18n/ro/CHANGELOG.mdindex 710126e55d..0eebb599cc 100644--- a/docs/i18n/ro/CHANGELOG.md+++ b/docs/i18n/ro/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Securitate+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentație+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentație+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Securitate+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Securitate+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentație+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Securitate+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentație+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Securitate+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcționalități+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Securitate+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcționalități+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcționalități+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Securitate+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Securitate+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentație+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentație+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcționalități+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcționalități+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcționalități+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Securitate+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcționalități+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcționalități+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcționalități+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcționalități+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcționalități+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcționalități+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcționalități+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcționalități+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Securitate+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcționalități+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcționalități+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcționalități+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcționalități+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcționalități+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcționalități+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentație+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcționalități+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/ru/CHANGELOG.md b/docs/i18n/ru/CHANGELOG.mdindex 07ecec92b6..2c3ed73b0c 100644--- a/docs/i18n/ru/CHANGELOG.md+++ b/docs/i18n/ru/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Безопасность+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Документация+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Документация+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Безопасность+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Безопасность+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Документация+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Безопасность+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Документация+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Безопасность+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Возможности+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Безопасность+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Возможности+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Возможности+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Безопасность+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Безопасность+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Документация+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Документация+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Возможности+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Возможности+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Возможности+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Безопасность+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Возможности+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Возможности+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Возможности+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Возможности+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Возможности+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Возможности+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Возможности+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Возможности+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Безопасность+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Возможности+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Возможности+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Возможности+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Возможности+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Возможности+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Возможности+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Документация+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Возможности+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/sk/CHANGELOG.md b/docs/i18n/sk/CHANGELOG.mdindex ded38fac22..a015ede846 100644--- a/docs/i18n/sk/CHANGELOG.md+++ b/docs/i18n/sk/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Bezpečnosť+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentácia+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentácia+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Bezpečnosť+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Bezpečnosť+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentácia+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Bezpečnosť+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentácia+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Bezpečnosť+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funkcie+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Bezpečnosť+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funkcie+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funkcie+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Bezpečnosť+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Bezpečnosť+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentácia+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentácia+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funkcie+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funkcie+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funkcie+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Bezpečnosť+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funkcie+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funkcie+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funkcie+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funkcie+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funkcie+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funkcie+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funkcie+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funkcie+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Bezpečnosť+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funkcie+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funkcie+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funkcie+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funkcie+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funkcie+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funkcie+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentácia+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funkcie+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/sv/CHANGELOG.md b/docs/i18n/sv/CHANGELOG.mdindex c53c908d55..d3a306af64 100644--- a/docs/i18n/sv/CHANGELOG.md+++ b/docs/i18n/sv/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Säkerhet+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Dokumentation+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Dokumentation+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Säkerhet+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Säkerhet+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Dokumentation+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Säkerhet+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Dokumentation+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Säkerhet+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funktioner+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Säkerhet+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funktioner+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funktioner+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Säkerhet+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Säkerhet+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Dokumentation+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Dokumentation+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funktioner+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funktioner+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funktioner+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Säkerhet+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funktioner+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funktioner+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funktioner+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funktioner+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funktioner+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funktioner+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funktioner+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funktioner+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Säkerhet+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funktioner+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funktioner+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funktioner+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funktioner+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funktioner+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funktioner+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Dokumentation+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funktioner+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/sw/CHANGELOG.md b/docs/i18n/sw/CHANGELOG.mdindex 5646552e82..4ee30fd93b 100644--- a/docs/i18n/sw/CHANGELOG.md+++ b/docs/i18n/sw/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/ta/CHANGELOG.md b/docs/i18n/ta/CHANGELOG.mdindex 169bbe6a59..7170f741f1 100644--- a/docs/i18n/ta/CHANGELOG.md+++ b/docs/i18n/ta/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/te/CHANGELOG.md b/docs/i18n/te/CHANGELOG.mdindex 11f9eca138..13c6202fdb 100644--- a/docs/i18n/te/CHANGELOG.md+++ b/docs/i18n/te/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/th/CHANGELOG.md b/docs/i18n/th/CHANGELOG.mdindex 2e365bc8fb..9e36533ab2 100644--- a/docs/i18n/th/CHANGELOG.md+++ b/docs/i18n/th/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### ความปลอดภัย+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### เอกสาร+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### เอกสาร+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### ความปลอดภัย+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### ความปลอดภัย+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### เอกสาร+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### ความปลอดภัย+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### เอกสาร+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### ความปลอดภัย+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### คุณสมบัติ+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### ความปลอดภัย+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### คุณสมบัติ+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### คุณสมบัติ+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### ความปลอดภัย+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### ความปลอดภัย+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### เอกสาร+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### เอกสาร+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### คุณสมบัติ+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### คุณสมบัติ+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### คุณสมบัติ+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### ความปลอดภัย+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### คุณสมบัติ+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### คุณสมบัติ+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### คุณสมบัติ+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### คุณสมบัติ+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### คุณสมบัติ+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### คุณสมบัติ+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### คุณสมบัติ+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### คุณสมบัติ+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### ความปลอดภัย+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### คุณสมบัติ+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### คุณสมบัติ+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### คุณสมบัติ+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### คุณสมบัติ+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### คุณสมบัติ+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### คุณสมบัติ+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### เอกสาร+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### คุณสมบัติ+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/tr/CHANGELOG.md b/docs/i18n/tr/CHANGELOG.mdindex c9fbad6094..2f0a42b12a 100644--- a/docs/i18n/tr/CHANGELOG.md+++ b/docs/i18n/tr/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Güvenlik+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Belgeler+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Belgeler+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Güvenlik+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Güvenlik+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Belgeler+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Güvenlik+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Belgeler+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Güvenlik+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Özellikler+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Güvenlik+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Özellikler+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Özellikler+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Güvenlik+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Güvenlik+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Belgeler+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Belgeler+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Özellikler+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Özellikler+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Özellikler+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Güvenlik+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Özellikler+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Özellikler+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Özellikler+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Özellikler+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Özellikler+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Özellikler+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Özellikler+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Özellikler+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Güvenlik+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Özellikler+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Özellikler+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Özellikler+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Özellikler+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Özellikler+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Özellikler+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Belgeler+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Özellikler+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/uk-UA/CHANGELOG.md b/docs/i18n/uk-UA/CHANGELOG.mdindex 3cfb253e67..52a3e84167 100644--- a/docs/i18n/uk-UA/CHANGELOG.md+++ b/docs/i18n/uk-UA/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Безпека+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Документація+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Документація+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Безпека+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Безпека+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Документація+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Безпека+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Документація+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Безпека+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Можливості+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Безпека+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Можливості+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Можливості+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Безпека+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Безпека+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Документація+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Документація+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Можливості+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Можливості+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Можливості+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Безпека+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Можливості+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Можливості+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Можливості+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Можливості+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Можливості+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Можливості+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Можливості+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Можливості+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Безпека+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Можливості+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Можливості+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Можливості+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Можливості+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Можливості+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Можливості+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Документація+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Можливості+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/ur/CHANGELOG.md b/docs/i18n/ur/CHANGELOG.mdindex 1a4c56bb07..5284e97910 100644--- a/docs/i18n/ur/CHANGELOG.md+++ b/docs/i18n/ur/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Seguridad+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Documentación+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Documentación+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Seguridad+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Seguridad+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Documentación+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Seguridad+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Documentación+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Seguridad+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Funcionalidades+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Seguridad+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Funcionalidades+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Funcionalidades+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Seguridad+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Seguridad+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Documentación+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Documentación+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Funcionalidades+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Funcionalidades+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Funcionalidades+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Seguridad+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Funcionalidades+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Funcionalidades+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Funcionalidades+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Funcionalidades+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Funcionalidades+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Funcionalidades+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Funcionalidades+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Seguridad+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Funcionalidades+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Funcionalidades+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Funcionalidades+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Funcionalidades+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Funcionalidades+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Funcionalidades+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Documentación+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Funcionalidades+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/vi/CHANGELOG.md b/docs/i18n/vi/CHANGELOG.mdindex 98212107b0..b04bcf78cb 100644--- a/docs/i18n/vi/CHANGELOG.md+++ b/docs/i18n/vi/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### Bảo mật+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### Tài liệu+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### Tài liệu+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### Bảo mật+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### Bảo mật+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### Tài liệu+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### Bảo mật+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### Tài liệu+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### Bảo mật+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### Tính năng+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### Bảo mật+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### Tính năng+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### Tính năng+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### Bảo mật+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### Bảo mật+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### Tài liệu+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### Tài liệu+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### Tính năng+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### Tính năng+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### Tính năng+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### Bảo mật+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### Tính năng+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### Tính năng+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### Tính năng+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### Tính năng+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### Tính năng+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### Tính năng+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### Tính năng+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### Tính năng+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### Bảo mật+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### Tính năng+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### Tính năng+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### Tính năng+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### Tính năng+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### Tính năng+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### Tính năng+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### Tài liệu+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### Tính năng+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/docs/i18n/zh-CN/CHANGELOG.md b/docs/i18n/zh-CN/CHANGELOG.mdindex 90629bd322..6c4580d768 100644--- a/docs/i18n/zh-CN/CHANGELOG.md+++ b/docs/i18n/zh-CN/CHANGELOG.md@@ -368,7 +368,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(proxy):** implement bulk proxy import via pipe-delimited parser with update-or-create (upsert) logic and real-time preview table- **feat(pwa):** add fullscreen installable PWA with manifest, service worker, and cross-platform app icons (#1728)-### 安全+### 🔒 Security- **security:** replace insecure `Math.random` with `crypto.getRandomValues` for fallback UUID generation to resolve CodeQL CWE-338 finding (#182)@@ -533,7 +533,7 @@ We identified that **155 community PRs** across the entire project history (from- **fix(transport):** Prevent memory bloat and database exhaustion from large, fragmented streaming responses. Implemented `ByteQueue` in `kiro.ts` for zero-copy binary accumulation, refactored `antigravity.ts` for incremental SSE parsing, and enforced a strict 512KB tiered truncation limit (`MAX_CALL_LOG_ARTIFACT_BYTES`) on stream request logs and call artifacts (#1647).- **chore(ci):** Update build environment dependencies — bump Node to `24.15.0`, `actions/checkout@v6`, `docker/build-push-action@v7`, pin `actions/setup-python` to major tag (#1646 — thanks @backryun).-### 文档+### 📝 Documentation- **docs(env):** Add `OMNIROUTE_ALLOW_PRIVATE_PROVIDER_URLS` to `.env.example` with documentation for LM Studio and other local provider use cases (#1623).@@ -683,7 +683,7 @@ We identified that **155 community PRs** across the entire project history (from- **test(next):** Align transpile package expectations for the Next.js standalone build.- **test(ci):** Fix CI-only test failures from environment differences — clear `INITIAL_PASSWORD` and `JWT_SECRET` in integration tests, handle `XDG_CONFIG_HOME` for guide-settings tests.-### 文档+### 📚 Documentation- **docs:** Update the root changelog with all release-branch changes through 2026-04-24, including PRs #1544, #1555, #1551, #1550, #1548, #1547, #1541, #1538, #1536, and #1527.- **docs:** Fix broken README and localized documentation links. (#1536)@@ -825,7 +825,7 @@ We identified that **155 community PRs** across the entire project history (from- **feat(providers):** Expand image provider registry with extended model support including SD3.5, FLUX, and DALL-E 3 HD configurations- **feat(combos):** Add new routing strategies and full i18n support for agent features section across 31 languages-### 安全+### 🔒 Security- **security:** Resolve 18 GitHub CodeQL scan alerts including ReDoS, incomplete sanitization, and bad HTML filtering regexp patterns- **fix(auth):** Seal privilege escalation vector by enforcing JWT session checking exclusively on `/api/keys` management endpoints (#1353)@@ -1092,7 +1092,7 @@ We identified that **155 community PRs** across the entire project history (from- **Duplicate `auto` in Combo Strategy Schema:** Removed duplicate `"auto"` entry from `comboStrategySchema` (was listed on both line 104 and 108). Harmless to Zod runtime but cleaned up to avoid confusion. Schema now has exactly 13 unique strategy values- **Legacy Combo Refs Normalization:** Fixed combo step normalization to preserve legacy string combo references during CRUD operations, preventing data loss when editing combos created before the v2 step architecture-### 安全+### 🔒 Security- **Auth Bypass on Backup Routes (Critical):** Added `isAuthenticated` guards to `/api/db-backups/exportAll` (full database export) and `/api/db-backups` (list, create, and restore backups) — both were previously accessible without authentication- **Auth Guard on Translator Save:** Added `isAuthenticated` guard to `/api/translator/save` for defense-in-depth consistency@@ -1497,7 +1497,7 @@ We identified that **155 community PRs** across the entire project history (from- **Updated Sub-dependencies:** Bumped `hono` to `4.12.12` and `@hono/node-server` to `1.19.13` to patch critical security gaps (#1063, #1064, #1067, #1068).-### 文档+### 📚 Documentation- **Documentation Synchronization:** Updated system documentation (README, Architecture, Features, Tools, Troubleshooting) and synced `i18n` configurations to match the v3.5.5 context relay patterns and proxy troubleshooting steps.- **Context Relay Delivery Notes:** Documented the current architecture, runtime flow, and Codex-focused scope in the feature docs, changelog, and agent guidance.@@ -1573,7 +1573,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.5.3] - 2026-04-07-### 安全+### Security- **Vulnerabilities:** Fully remediated 12 High-Severity CodeQL vulnerabilities by migrating from Math.random to `crypto.randomUUID()`, wrapping SSE injection points with aggressive backslash escaping, sanitizing trailing HTTP fragments, and enforcing rigid SSRF HTTP verification schemes across internal routes.- **Dependencies:** Upgraded Next.js to `^16.2.2` and Vite to `>=8.0.5` resolving critical DoS, arbitrary file reads and CSRF vectors in the build/server environments.@@ -1589,7 +1589,7 @@ We identified that **155 community PRs** across the entire project history (from- **CI/CD Stabilization:** Prevented random GitHub Runner freezes by decoupling sharded processes, adjusting test concurrencies, unref-ing active connections on server teardown, and strictly capping job timeout durations.-### 文档+### Documentation- **I18n Engine:** Synchronized and pushed deep Machine Translation updates across all 32 natively-supported languages (682 translation nodes aligned).@@ -1750,7 +1750,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.8] — 2026-04-03-### 安全+### Security- Fully remediated all outstanding Github Advanced Security (CodeQL) findings and Dependabot alerts.- Fixed insecure randomness vulnerabilities by migrating from `Math.random` to `crypto.randomUUID()`.@@ -1762,7 +1762,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.7] — 2026-04-03-### 功能特点+### Features- Added `Cryptography` node to Monitoring and MCP health checks (#798)- Hardened model-catalog route permissions mapping (`/models`) (#781)@@ -1778,7 +1778,7 @@ We identified that **155 community PRs** across the entire project history (from- Fixed MCP standalone module-resolution (`ERR_MODULE_NOT_FOUND`) via `esbuild` (#936)- Fixed NVIDIA NIM routing credential resolution alias mismatch (#931)-### 安全+### Security- Added safe strict input boundary protection against raw `shell: true` remote-code execution injections.@@ -2025,7 +2025,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.4.0] - 2026-03-31-### 功能特点+### 🚀 Features- **Subscription Utilization Analytics:** Added quota snapshot time-series tracking, Provider Utilization and Combo Health tabs with recharts visualizations, and corresponding API endpoints (#847)- **SQLite Backup Control:** New `OMNIROUTE_DISABLE_AUTO_BACKUP` env flag to disable automatic SQLite backups (#846)@@ -2077,7 +2077,7 @@ We identified that **155 community PRs** across the entire project history (from## [3.3.8] - 2026-03-30-### 功能特点+### 🚀 Features- **Models API Filtering:** Endpoint `/v1/models` now dynamically filters its list based on the permissions tied to the `Authorization: Bearer <token>` when restricted access is on (#781)- **Qoder Integration:** Native integration for Qoder AI natively replacing the legacy iFlow platform mappings (#660)@@ -2903,7 +2903,7 @@ We identified that **155 community PRs** across the entire project history (from- **NaN tokens in Claude Code / client responses (#617):**- `sanitizeUsage()` now cross-maps `input_tokens`→`prompt_tokens` and `output_tokens`→`completion_tokens` before the whitelist filter, fixing responses showing NaN/0 token counts when providers return Claude-style usage field names-### 安全+### 🔒 Security- Updated `yaml` package to fix stack overflow vulnerability (GHSA-48c2-rrv3-qjmp)@@ -3579,7 +3579,7 @@ docker pull diegosouzapw/omniroute:3.0.0- **SVG fallback**: `ProviderIcon` component updated with 4-tier strategy: Lobehub → PNG → SVG → Generic icon- **Agents fingerprinting**: Synced with CLI tools — added droid, openclaw, copilot, opencode to fingerprint list (14 total)-### 安全+### 🔒 Security- **CVE fix**: Resolved dompurify XSS vulnerability (GHSA-v2wj-7wpq-c8vv) via npm overrides forcing `dompurify@^3.3.2`- `npm audit` now reports **0 vulnerabilities**@@ -3942,7 +3942,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **#510** — Windows: MSYS2/Git-Bash paths (`/c/Program Files/...`) are now normalized to `C:\Program Files\...`- **#492** — `omniroute` CLI now detects `mise`/`nvm` when `app/server.js` is missing and shows targeted fix-### 文档+### 📖 Documentation- **#513** — Docker password reset: `INITIAL_PASSWORD` env var workaround documented- **#520** — pnpm: `pnpm approve-builds better-sqlite3` documented@@ -4039,7 +4039,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **feat(executors/cloudflare-ai)**: New `CloudflareAIExecutor` — dynamic URL construction requires `accountId` in provider credentials- **feat(executors)**: Register `pollinations`, `pol`, `cloudflare-ai`, `cf` executor mappings-### 文档+### 📝 Documentation- **docs(readme)**: Expanded free combo stack to 11 providers ($0 forever)- **docs(readme)**: Added 4 new free provider sections (LongCat, Pollinations, Cloudflare AI, Scaleway) with model tables@@ -4191,7 +4191,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **OAuth batch test crash** (ERR_CONNECTION_REFUSED): Replaced sequential for-loop with 5-connection concurrency limit + 30s per-connection timeout via `Promise.race()` + `Promise.allSettled()`. Prevents server crash when testing large OAuth provider groups (~30+ connections).-### 功能特点+### Features- **"Test All" button on provider pages**: Individual provider pages (e.g., `/providers/codex`) now show a "Test All" button in the Connections header when there are 2+ connections. Uses `POST /api/providers/test-batch` with `{mode: "provider", providerId}`. Results displayed in a modal with pass/fail summary and per-connection diagnosis.@@ -4219,7 +4219,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Merge PR #494 (MiniMax role fix), fix KIRO MITM dashboard, triage 8 issues.-### 功能特点+### Features- **MiniMax developer→system role fix** (PR #494 by @zhangqiang8vip): Per-model `preserveDeveloperRole` toggle. Adds "Compatibility" UI in providers page. Fixes 422 "role param error" for MiniMax and similar gateways.- **roleNormalizer**: `normalizeDeveloperRole()` now accepts `preserveDeveloperRole` parameter with tri-state behavior (undefined=keep, true=keep, false=convert).@@ -4275,7 +4275,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Gemini CLI deprecation, VM guide i18n fix, dependabot security fix, provider schema expansion.-### 功能特点+### Features- **Gemini CLI Deprecation** (#462): Mark `gemini-cli` provider as deprecated with warning — Google restricts third-party OAuth usage from March 2026- **Provider Schema** (#462): Expand Zod validation with `deprecated`, `deprecationReason`, `hasFree`, `freeNote`, `authHint`, `apiHint` optional fields@@ -4284,7 +4284,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **VM Guide i18n** (#471): Add `VM_DEPLOYMENT_GUIDE.md` to i18n translation pipeline, regenerate all 30 locale translations from English source (were stuck in Portuguese)-### 安全+### Security- **deps**: Bump `flatted` 3.3.3 → 3.4.2 — fixes CWE-1321 prototype pollution (#484, @dependabot)@@ -4304,7 +4304,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Czech i18n, SSE protocol fix, VM guide translation.-### 功能特点+### Features- **Czech Language** (#482): Full Czech (cs) i18n — 22 docs, 2606 UI strings, language switcher updates (@zen0bit)- **VM Deployment Guide**: Translated from Portuguese to English as the source document (@zen0bit)@@ -4323,7 +4323,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: 2 merged PRs, model aliases routing fix, log export, and issue triage.-### 功能特点+### Features- **Log Export**: New Export button on `/dashboard/logs` with time range dropdown (1h, 6h, 12h, 24h). Downloads JSON of request/proxy/call logs via `/api/logs/export` API (#user-request)@@ -4343,7 +4343,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Five community PRs — streaming call log fixes, Kiro compatibility, cache token analytics, Chinese translation, and configurable tool call IDs.-### 功能特点+### ✨ Features- **feat(logs)**: Call log response content now correctly accumulated from raw provider chunks (OpenAI/Claude/Gemini) before translation, fixing empty response payloads in streaming mode (#470, @zhangqiang8vip)- **feat(providers)**: Per-model configurable 9-char tool call ID normalization (Mistral-style) — only models with the option enabled get truncated IDs (#470)@@ -4373,7 +4373,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Bailian Coding Plan provider with editable base URLs, plus community contributions for Alibaba Cloud and Kimi Coding.-### 功能特点+### ✨ Features- **feat(providers)**: Added Bailian Coding Plan (`bailian-coding-plan`) — Alibaba Model Studio with Anthropic-compatible API. Static catalog of 8 models including Qwen3.5 Plus, Qwen3 Coder, MiniMax M2.5, GLM 5, and Kimi K2.5. Includes custom auth validation (400=valid, 401/403=invalid) (#467, @Mind-Dragon)- **feat(admin)**: Editable default URL in Provider Admin create/edit flows — users can configure custom base URLs per connection. Persisted in `providerSpecificData.baseUrl` with Zod schema validation rejecting non-http(s) schemes (#467)@@ -4388,7 +4388,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Two new community-contributed providers (Alibaba Cloud Coding, Kimi Coding API-key) and Docker pino fix.-### 功能特点+### ✨ Features- **feat(providers)**: Added Alibaba Cloud Coding Plan support with two OpenAI-compatible endpoints — `alicode` (China) and `alicode-intl` (International), each with 8 models (#465, @dtk1985)- **feat(providers)**: Added dedicated `kimi-coding-apikey` provider path — API-key-based Kimi Coding access is no longer forced through OAuth-only `kimi-coding` route. Includes registry, constants, models API, config, and validation test (#463, @Mind-Dragon)@@ -4413,7 +4413,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Codex responses subpath passthrough natively supported, Windows MITM crash fixed, and Combos agent schemas adjusted.-### 功能特点+### ✨ Features- **feat(codex)**: Native responses subpath passthrough for Codex — natively routes `POST /v1/responses/compact` to Codex upstream, maintaining Claude Code compatibility without stripping the `/compact` suffix (#457)@@ -4453,7 +4453,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(budget)**: "Save Limits" no longer returns 422 — `warningThreshold` is now correctly sent as fraction (0–1) instead of percentage (0–100) (#451)- **fix(combos)**: `<omniModel>` internal cache tag is now stripped before forwarding requests to providers, preventing cache session breaks (#454)-### 功能特点+### ✨ Features- **feat(combos)**: Agent Features section added to combo create/edit modal — expose `system_message` override, `tool_filter_regex`, and `context_cache_protection` directly from the dashboard (#454)@@ -4509,7 +4509,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: Search Tools dashboard, i18n fixes, Copilot limits, Serper validation fix.-### 功能特点+### 🚀 Features- **feat(search)**: Add Search Playground (10th endpoint), Search Tools page with Compare Providers/Rerank Pipeline/Search History, local rerank routing, auth guards on search API (#443 by @Regis-RCR)- New route: `/dashboard/search-tools`@@ -4609,7 +4609,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- DB migration: `request_type` column on `call_logs` for non-chat request tracking- Zod validation (`v1SearchSchema`), auth-gated, cost recorded via `recordCost()`-### 安全+### 🔒 Security- **deps**: Next.js 16.1.6 → 16.1.7 — fixes 6 CVEs:- **Critical**: CVE-2026-29057 (HTTP request smuggling via http-proxy)@@ -4739,7 +4739,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **005_combo_agent_fields.sql**: `ALTER TABLE combos ADD COLUMN system_message TEXT DEFAULT NULL`, `tool_filter_regex TEXT DEFAULT NULL`, `context_cache_protection INTEGER DEFAULT 0`- **006_detailed_request_logs.sql**: New `request_detail_logs` table with 500-entry ring-buffer trigger, opt-in via settings toggle-### 功能特点+### ✨ Features- **feat(combo)**: System Message Override per Combo (#399 — `system_message` field replaces or injects system prompt before forwarding to provider)- **feat(combo)**: Tool Filter Regex per Combo (#399 — `tool_filter_regex` keeps only tools matching pattern; supports OpenAI + Anthropic formats)@@ -4754,7 +4754,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> Sprint: SSE improvements, local provider_nodes extensions, proxy registry, Claude passthrough fixes.-### 功能特点+### ✨ Features- **feat(health)**: Background health check for local `provider_nodes` with exponential backoff (30s→300s) and `Promise.allSettled` to avoid blocking (#423, @Regis-RCR)- **feat(embeddings)**: Route `/v1/embeddings` to local `provider_nodes` — `buildDynamicEmbeddingProvider()` with hostname validation (#422, @Regis-RCR)@@ -4896,7 +4896,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Turbopack hash-strip now runs at **compile time** for ALL packages — not just `better-sqlite3`. Step 5.6 in `prepublish.mjs` walks every `.js` in `app/.next/server/` and strips the 16-char hex suffix from any hashed `require()`. Fixes `zod-dcb22c...`, `pino-...`, etc. MODULE_NOT_FOUND on global npm installs. Closes #398- **fix(deploy)**: PM2 on both VPS was pointing to stale git-clone directories. Reconfigured to `app/server.js` in the npm global package. Updated `/deploy-vps` workflow to use `npm pack + scp` (npm registry rejects 299MB packages).-### 功能特点+### ✨ Features- **feat(provider)**: Synthetic ([synthetic.new](https://synthetic.new)) — privacy-focused OpenAI-compatible inference. `passthroughModels: true` for dynamic HuggingFace model catalog. Initial models: Kimi K2.5, MiniMax M2.5, GLM 4.7, DeepSeek V3.2. (PR #404 by @Regis-RCR)@@ -4926,7 +4926,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(build)**: Extended webpack `externals` hash-strip to cover ALL `serverExternalPackages`, not just `better-sqlite3`. Next.js 16 Turbopack hashes `zod`, `pino`, and every other server-external package into names like `zod-dcb22c6336e0bc69` that don't exist in `node_modules` at runtime. A HASH_PATTERN regex catch-all now strips the 16-char suffix and falls back to the base package name. Also added `NEXT_PRIVATE_BUILD_WORKER=0` in `prepublish.mjs` to reinforce webpack mode, plus a post-build scan that reports any remaining hashed refs. (#396, #398, PR #403)- **fix(chat)**: Anthropic-format tool names (`tool.name` without `.function` wrapper) were silently dropped by the empty-name filter introduced in #346. LiteLLM proxies requests with `anthropic/` prefix in Anthropic Messages API format, causing all tools to be filtered and Anthropic to return `400: tool_choice.any may only be specified while providing tools`. Fixed by falling back to `tool.name` when `tool.function.name` is absent. Added 8 regression unit tests. (PR #397)-### 功能特点+### ✨ Features- **feat(api)**: Custom endpoint paths for OpenAI-compatible provider nodes — configure `chatPath` and `modelsPath` per node (e.g. `/v4/chat/completions`) in the provider connection UI. Includes a DB migration (`003_provider_node_custom_paths.sql`) and URL path sanitization (no `..` traversal, must start with `/`). (PR #400)- **feat(provider)**: Alibaba Cloud DashScope added as OpenAI-compatible provider. International endpoint: `dashscope-intl.aliyuncs.com/compatible-mode/v1`. 12 models: `qwen-max`, `qwen-plus`, `qwen-turbo`, `qwen3-coder-plus/flash`, `qwq-plus`, `qwq-32b`, `qwen3-32b`, `qwen3-235b-a22b`. Auth: Bearer API key.@@ -4985,7 +4985,7 @@ OmniRoute now automatically refreshes model lists for connected providers every- **fix(oauth)**: Qoder (and other providers that redirect to their own UI) no longer leave the OAuth modal stuck at "Waiting for Authorization" — popup-closed detector auto-transitions to manual URL input mode (#344)- **fix(logs)**: Request log table is now readable in light mode — status badges, token counts, and combo tags use adaptive `dark:` color classes (#378)-### 功能特点+### ✨ Features- **feat(kiro)**: Kiro credit tracking added to usage fetcher — queries `getUserCredits` from AWS CodeWhisperer endpoint (#337)@@ -5386,7 +5386,7 @@ OmniRoute now automatically refreshes model lists for connected providers every> **Major release** — Free Stack ecosystem, transcription playground overhaul, 44+ providers, comprehensive free tier documentation, and UI improvements across the board.-### 功能特点+### ✨ Features- **Combos: Free Stack template** — New 4th template "Free Stack ($0)" using round-robin across Kiro + Qoder + Qwen + Gemini CLI. Suggests the pre-built zero-cost combo on first use.- **Media/Transcription: Deepgram as default** — Deepgram (Nova 3, $200 free) is now the default transcription provider. AssemblyAI ($50 free) and Groq Whisper (free forever) shown with free credit badges.@@ -5397,7 +5397,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.16] - 2026-03-13-### 文档+### 📖 Documentation- **README: 44+ Providers** — Updated all 3 occurrences of "36+ providers" to "44+" reflecting the actual codebase count (44 providers in providers.ts)- **README: New Section "🆓 Free Models — What You Actually Get"** — Added 7-provider table with per-model rate limits for: Kiro (Claude unlimited via AWS Builder ID), Qoder (5 models unlimited), Qwen (4 models unlimited), Gemini CLI (180K/mo), NVIDIA NIM (~40 RPM dev-forever), Cerebras (1M tok/day / 60K TPM), Groq (30 RPM / 14.4K RPD). Includes the \/usr/bin/bash Ultimate Free Stack combo recommendation.@@ -5407,7 +5407,7 @@ OmniRoute now automatically refreshes model lists for connected providers every## [2.3.15] - 2026-03-13-### 功能特点+### ✨ Features- **Auto-Combo Dashboard (Tier Priority)**: Added `🏷️ Tier` as the 7th scoring factor label in the `/dashboard/auto-combo` factor breakdown display — all 7 Auto-Combo scoring factors are now visible.- **i18n — autoCombo section**: Added 20 new translation keys for the Auto-Combo dashboard (`title`, `status`, `modePack`, `providerScores`, `factorTierPriority`, etc.) to all 30 language files.diff --git a/src/i18n/messages/zh-CN.json b/src/i18n/messages/zh-CN.jsonindex 7df46e9a68..83e5909a72 100644--- a/src/i18n/messages/zh-CN.json+++ b/src/i18n/messages/zh-CN.json@@ -2493,6 +2493,9 @@"testKeyPlaceholder": "sk-...(仅用于验证)","providerNotFound": "未找到提供商","deleteConnectionConfirm": "删除这个连接吗?",+ "batchDeleteSelected": "Delete Selected ({count})",+ "batchDeleteConfirm": "Delete {count} connection(s)? This action cannot be undone.",+ "batchDeleteSuccess": "Deleted {count} connection(s)","failedSetAlias": "设置别名失败","failedSaveConnection": "保存连接失败","failedSaveConnectionRetry": "无法保存连接。请再试一次。",@@ -3121,6 +3124,8 @@"cliFingerprintEnabled": "已有 {count} 个提供商启用了 CLI 指纹","enableFingerprintTitle": "为 {provider} 启用指纹","disableFingerprintTitle": "为 {provider} 禁用指纹",+ "forcedFingerprintTitle": "{provider} 始终启用 — OAuth 账户安全所必需;无法关闭。",+ "forcedFingerprintBadge": "必需","routingStrategy": "路由策略","routingAdvancedGuideTitle": "高级路由指南","routingAdvancedGuideHint1": "需要可预测优先级时使用 Fill First,需要公平分配时使用 Round Robin,需要延迟弹性时使用 P2C。",@@ -5300,4 +5305,4 @@"no": "否","noTargets": "未配置目标路由。"}-}+}\ No newline at end of file