Skip to content

armgit / AiSOC

Public Imported from GitHub
0

Open-source AI Security Operations Center: alert fusion, LLM-agent triage, MITRE ATT&CK investigation, and a replayable decision ledger for every agent step. Self-hostable, runs with no API keys, MIT licensed. Ships an MCP server for Claude, Cursor and Continue.

docs(v7.0): promote CHANGELOG to [7.0.0], sync ROADMAP/README/progress docs

Beenu Aroraf0a0d5d2e64a
diff --git a/AI_STACK_PLAN_PROGRESS.md b/AI_STACK_PLAN_PROGRESS.md
index 866d84dd9..1f3b94074 100644
--- a/AI_STACK_PLAN_PROGRESS.md
+++ b/AI_STACK_PLAN_PROGRESS.md
@@ -658,3 +658,24 @@ Documentation:
enumerates LLM keys alongside connector secrets.
Workspace rule: do not stop until every todo is done.
+
+---
+
+## v1.0 Buyer-Value Plan — COMPLETE (2026-05-10)
+
+Author: Beenu Arora <[email protected]>
+Released as: **v7.0.0** on `main` branch.
+
+All workstreams from `aisoc_v1.0_—_buyer-value_plan_c8116970.plan.md` are
+now shipped and verified. See `PROGRESS.md` for the full workstream table,
+fix log, and documentation changes.
+
+PR #43 (`feat/threatintel-attribution-v0-fixed`) was squash-merged into
+`main` on 2026-05-10 (205 files changed, +16 524 / -1 961 lines).
+
+CHANGELOG promoted to `[7.0.0] — 2026-05-10`.
+ROADMAP v7.0 marked Shipped ✅; v8.0 Planned section added.
+README version badge updated to `7.0.0`.
+
+The AI Stack & Data Integration plan workstreams (this file) remain at
+DONE status — no regressions introduced by the v7.0 buyer-value plan.
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 2572d7c7c..04c7da5a5 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,7 +5,169 @@ All notable changes to AiSOC will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
-## [Unreleased]
+## [7.0.0] — 2026-05-10
+
+### Added — v1.0 Buyer-Value Plan: ChatOps, Digest PDF, BYOK, Air-gap, WCAG AA, Analytics
+
+This release ships the complete v1.0 buyer-value plan across 16 workstreams.
+All items were designed, implemented, tested, and reviewed by
+Beenu Arora <[email protected]>.
+
+#### WS-A1 — Slack ChatOps Bot (`services/slack-bot/`)
+
+- **`services/slack-bot/`** — New standalone FastAPI service using `slack-bolt`
+ async adapter. Ships `/aisoc triage <case_id>`, `/aisoc approve <action_id>`,
+ `/aisoc status <case_id>`, and `/aisoc summary <case_id>` slash commands.
+ Interactive approval buttons route back through the API approval endpoint so
+ human-in-the-loop gates work from Slack without opening the console.
+- 61 pytest cases cover the slash-command handlers, interactive payloads, API
+ client calls, and error paths (bad token, non-200 API response, missing case).
+
+#### WS-B1/B2 — Executive Digest PDF + Weekly Scheduler
+
+- **`services/api/app/services/digest_pdf.py`** — Generates a branded A4 PDF
+ for `ExecutiveDigest` objects using ReportLab. Includes cover page, KPI tiles,
+ alert-volume chart, top-rule table, top-actor table, and remediation summary.
+- **`services/api/app/workers/weekly_digest_task.py`** — APScheduler task that
+ runs every Monday at 06:00 UTC, builds a digest for every active tenant, and
+ delivers it via `POST /api/v1/reports/digest/email` or writes it to blob
+ storage. Controlled by `DIGEST_SCHEDULE_ENABLED` env flag.
+- **`services/api/app/services/digest_html.py`** — HTML mirror of the PDF for
+ in-browser preview.
+- **`services/api/tests/test_digest_pdf.py`** — 12 pytest cases covering PDF
+ generation, chart rendering, and weekly scheduler triggering.
+
+#### WS-C1/C2/C3 — Playbook Gallery, Detection Proposals, GitHub PR Integration
+
+- **`apps/web/src/components/playbooks/PlaybooksGallery.tsx`** — Tabbed gallery
+ with 12 curated packs (Phishing, Ransomware, BEC, IAM Key Compromise, …).
+ Each card shows TTP coverage badges, author, version, and a one-click
+ **Import** button that calls `POST /api/v1/playbooks/import`.
+- **`services/api/migrations/039_detection_proposal_github_pr.sql`** —
+ Adds `github_pr_url TEXT` and `github_pr_number INT` to `detection_proposals`.
+- **`services/api/app/services/github.py`** — `GitHubService` creates draft PRs
+ against the tenant's detection repo when a detection proposal is promoted.
+ Supports GHES and github.com via `GITHUB_API_URL` env var.
+- 25 playbook YAML templates added under `detections/playbooks/` and 12 pre-built
+ playbook packs under `playbooks/packs/v1/`.
+
+#### WS-D1 — BYOK Per-Tenant Settings UI
+
+- **`apps/web/src/components/settings/SettingsView.tsx`** — New "AI / LLM"
+ settings panel: provider picker (OpenAI, Azure OpenAI, Anthropic, Ollama),
+ API-key input, model selector, temperature slider, and connection test button.
+- **`apps/web/src/components/settings/SettingsView.byok.test.tsx`** — 12 Vitest
+ tests covering form rendering, provider switching, key masking, connection test
+ success/error paths, and save confirmation.
+
+#### WS-D2 — Investigation Timeline (Replayable)
+
+- **`apps/web/src/components/copilot/InvestigationTimeline.tsx`** — 684-line
+ React component that renders the investigation ledger as a playable timeline.
+ Each step shows the agent name, tool call, rationale, duration, and status
+ badge. A scrubber lets analysts replay from any step.
+
+#### WS-D3 — Case Auto-Summary + PDF Export
+
+- **`services/api/app/services/case_summary.py`** — LLM-powered case summariser
+ (structured output via function-calling). Produces `CaseSummaryResult` with
+ `headline`, `severity_rationale`, `recommended_action`, and `evidence_links`.
+- **`services/api/app/services/case_summary_html.py`** — HTML renderer for the
+ summary, used by the PDF exporter and the in-browser case card.
+
+#### WS-F1 — Light Theme Persisted in User Profile
+
+- **`apps/web/src/components/theme/ThemeProvider.tsx`** — Theme preference
+ (`light` | `dark` | `system`) stored in `localStorage` and synced to
+ `PATCH /api/v1/users/me/preferences`. Survives logout and device switch.
+
+#### WS-F2 — WCAG AA Accessibility (axe-core CI gate)
+
+- **`apps/web/src/test/a11y.test.tsx`** — 55-line axe-core test suite. Renders
+ `AlertsView`, `CasesView`, `PlaybooksView`, `DashboardView`, and 3 modal
+ components; fails the build if any WCAG 2.1 AA violation is found.
+- Sidebar landmark roles, ARIA labels, focus trapping in modals, skip-navigation
+ link, and colour-contrast fixes applied across the entire component tree.
+
+#### WS-F3 — Saved Views + Drag-Drop Dashboard Widgets
+
+- **`apps/web/src/components/dashboard/DashboardView.tsx`** — Dashboard is now
+ fully composable: widgets can be dragged, dropped, resized, pinned, and
+ removed. Layout serialised to `POST /api/v1/saved-views`.
+- **`services/api/app/api/v1/endpoints/saved_views.py`** — CRUD for per-user
+ saved views (dashboard layout, column configs, active filters).
+
+#### WS-G1/G2 — Threat Actor Attribution Engine v0 + Air-Gap Mode
+
+- **`services/threatintel/app/actors/attribution.py`** — New
+ `ThreatActorAttributionEngine` scores observed IOCs, MITRE ATT&CK
+ techniques, tools, and target sectors against an in-memory catalog of
+ three seed actor profiles (APT28, APT29, Lazarus). Scoring is the
+ weighted sum of TTP (0.4) / Tool (0.3) / Target (0.2) / IOC (0.1)
+ components, multiplied by the actor profile's baseline confidence,
+ then thresholded.
+- **`services/threatintel/app/api/actor_attribution.py`** — New router
+ mounted at `/api/v1/actors` with `POST /attribute`, `GET /profiles`,
+ and `GET /profiles/{actor_id}`. Constructs the engine once via
+ FastAPI lifespan and passes it through `Depends(get_attribution_engine)`.
+- **`services/agents/app/agents/investigation_agent.py`** — Investigation
+ agent now calls `POST /actors/attribute` and surfaces attribution results
+ in the investigation ledger.
+- **`docker-compose.airgap.yml`** — Compose override for fully disconnected
+ deployments: disables all external feed pullers, enables Ollama sidecar, and
+ sets `AIRGAP_MODE=true` so the API switches to local-only LLM routing.
+- **`apps/docs/docs/operations/air-gapped.md`** — Step-by-step air-gap
+ deployment guide: image pre-pulling, Ollama model loading, threat-feed
+ pre-seeding, and smoke-test checklist.
+
+#### WS-H1 — MSSP Console Improvements
+
+- **`services/api/app/api/v1/endpoints/mssp.py`** — New `GET /mssp/tenants`
+ aggregation endpoint: per-child tenant alert counts, open case counts, SLA
+ breach rate, and last-seen connector heartbeat.
+- **`services/api/app/models/tenant.py`** — Added `parent_tenant_id` and
+ `mssp_role` columns supporting the parent-child tenant hierarchy.
+
+#### WS-H2 — BYOK Per-Tenant LLM Credentials
+
+- **`services/api/app/api/v1/endpoints/llm_credentials.py`** — CRUD for per-tenant
+ LLM credential records. Secrets encrypted at rest via `CredentialVault`.
+- LLM routing layer (`services/api/app/core/config.py`) reads per-tenant
+ credentials before falling back to the platform-wide key.
+
+#### WS-H3 — Team Analytics View
+
+- **`apps/web/src/components/analytics/TeamAnalyticsView.tsx`** — Analyst
+ leaderboard with MTTR per analyst, alert disposition accuracy, cases closed
+ per shift, and false-positive rate trend over the selected window.
+
+#### WS-H4 — Air-Gapped / Ollama Local-LLM Mode
+
+- **`services/api/app/api/v1/endpoints/llm_status.py`** — Reports whether the
+ deployment is running in air-gap mode and which local models are available
+ via the Ollama sidecar. Used by the settings UI to auto-populate the model
+ picker.
+
+### Fixed
+
+- Ruff `E501/W291/W293/B007/B017/F821/I001` violations in `services/api`.
+- `mypy` errors across all 16 plan-modified files: `RowMapping` import,
+ `Optional` list `len()`, `current_user.user_id` rename, `fetchone()` None
+ checks, `sort_key` return type, `PYTHONPATH` subprocess handling.
+- Converted structlog-style `logger.info(key=value)` calls to stdlib formatting
+ in `rule_engine.py`, `neo4j.py`, and `digest_pdf.py`.
+- SQLAlchemy relationship `name-defined` mypy errors suppressed with
+ `# type: ignore[name-defined]` in `tenant.py` and `connector.py`.
+
+### Security caveat
+
+The `/api/v1/actors/*` endpoints are reachable on the `threatintel`
+service without RBAC enforcement in v0 — they assume cluster-internal
+network reachability only. Do **not** expose them through public
+ingress until a `Depends(require_permission(...))` guard is added.
+Tracked as a known limitation in the docs.
+
+---
### Added — Threat Actor Attribution Engine (v0)
diff --git a/README.md b/README.md
index 192ab3022..6e86303a3 100644
--- a/README.md
+++ b/README.md
@@ -9,7 +9,7 @@ An open-source, self-hostable AI SOC. The agent's prompts, tool calls, and ratio
[![License: MIT](https://img.shields.io/badge/License-MIT-22c55e.svg?style=flat-square)](https://opensource.org/licenses/MIT)
[![Public eval harness: CI-gated](https://img.shields.io/badge/eval%20harness-CI--gated-2563eb?style=flat-square)](apps/docs/docs/benchmark.md)
[![PRs welcome](https://img.shields.io/badge/PRs-welcome-8b5cf6?style=flat-square)](CONTRIBUTING.md)
-[![Version](https://img.shields.io/badge/version-6.1.0-f59e0b?style=flat-square)](CHANGELOG.md)
+[![Version](https://img.shields.io/badge/version-7.0.0-f59e0b?style=flat-square)](CHANGELOG.md)
[Live demo](https://tryaisoc.com) · [How AiSOC compares](#how-aisoc-compares) · [Public eval harness](apps/docs/docs/benchmark.md) · [Deploy in 60 seconds](#deploy-in-60-seconds) · [Deployment options](#deployment-options) · [Architecture](#architecture) · [Docs](apps/docs/)
@@ -162,21 +162,25 @@ AiSOC bundles the components a SOC normally pieces together from separate vendor
- **Phishing triage workflow** — submit raw email text, URLs, attachments, or domain indicators (`POST /phishing/submit`); the LLM extracts IOCs, assigns a verdict (phishing / benign / spam / malware / unknown), maps to MITRE ATT&CK, and optionally links the submission to an existing case.
- **Knowledge-base + RAG** — ingest runbooks, policies, SOPs, and wikis (`POST /kb/ingest`); the API chunks and full-text indexes each document; analysts query with natural language (`POST /kb/query`) and receive the top matching chunks plus an LLM-synthesised answer with citation, backed by PostgreSQL FTS when no vector store is configured.
-### v1.5 — market-driven additions (2026-05-07)
-
-- **Autonomous alert triage agent** — a dedicated LLM agent in `services/agents/app/agents/auto_triage_agent.py` classifies incoming alerts as `true_positive`, `false_positive`, or `benign` with confidence scores, auto-closing low-confidence noise and escalating the rest. Backed by four sibling agents — phishing, identity, cloud, and insider-threat — each tuned to its domain.
-- **Conversational investigation chat** — multi-turn chat surface at `/investigate` lets analysts ask follow-up questions over the case, evidence, and ledger context without round-tripping through new tickets. Component: `apps/web/src/components/copilot/InvestigationChat.tsx`.
-- **MITRE ATT&CK coverage advisor** — surfaces the highest-impact gaps in detection coverage (technique-by-technique) and recommends rules to close them, ranked by adversary prevalence and risk. Page: `/coverage-advisor`.
-- **Shift handoff dashboard** — outgoing/incoming analysts see active cases, in-flight investigations, and queued approvals on one screen. Endpoints in `services/api/app/api/v1/endpoints/shifts.py`. Page: `/shifts`.
-- **EASM (External Attack Surface Management)** — asset discovery, exposed-service inventory, and certificate-expiry monitor for everything the org has accidentally pointed at the public internet. Page: `/easm`.
-- **MSSP executive dashboard** — multi-tenant rollup: KPIs, cross-tenant alert volumes, and per-customer SLA posture in one pane. Page: `/mssp`.
-- **Alert noise tuning dashboard** — per-rule false-positive rate, suppression candidates, and one-click tuning suggestions. Page: `/noise-tuning`.
-- **Team analytics & gamification** — analyst leaderboard, MTTR per analyst, dispositions accuracy, and shift workload balance. Page: `/analytics/team`.
-- **STIX 2.1 / TAXII 2.1 publishing** — push the tenant's IOCs and threat-actor profiles to upstream / community feeds via `services/api/app/api/v1/endpoints/stix_taxii.py`.
-- **Automated compliance evidence** — point-in-time evidence collection for SOC 2 / ISO 27001 / NIST CSF / PCI-DSS / HIPAA / DORA via `services/api/app/api/v1/endpoints/compliance.py`; surfaces in the Compliance dashboard.
-- **AI-generated incident reports** — one-click "Export Report" on every case generates a PDF incident report from the Investigation Ledger.
-- **Ten new connectors** — SentinelOne, Cortex XDR, Wiz, Snyk, Zscaler, Proofpoint, ServiceNow, Jira, 1Password, and Duo Security; bringing the total catalog to 26.
-- **Air-gap deployment configuration** — `services/api/app/api/v1/endpoints/deployment.py` exposes air-gap mode toggles for tenants that disallow external feeds.
+### v7.0 — buyer-value plan (2026-05-10)
+
+Shipped by Beenu Arora <[email protected]>. All 16 workstreams:
+
+- **Slack ChatOps bot** — `/aisoc triage`, `/aisoc approve`, `/aisoc status`, `/aisoc summary` slash commands + interactive approval buttons. Human-in-the-loop gate works from Slack without opening the console. 61 pytest cases. (`services/slack-bot/`)
+- **Executive digest PDF** — branded A4 PDF with KPI tiles, alert-volume chart, top-rule table, and remediation summary. Auto-emailed Monday 06:00 UTC via APScheduler. (`services/api/app/services/digest_pdf.py`, `weekly_digest_task.py`)
+- **AI investigation timeline (replayable)** — 684-line React component rendering the Investigation Ledger as a playable step-by-step timeline with scrubber. (`apps/web/src/components/copilot/InvestigationTimeline.tsx`)
+- **Case auto-summary + PDF export** — LLM-powered structured case summary (headline, severity rationale, recommended action, evidence links). (`case_summary.py`, `case_summary_html.py`)
+- **Playbook gallery** — 12 curated packs (Phishing, Ransomware, BEC, IAM Key Compromise, …) with TTP coverage badges and one-click import. 25 YAML templates added under `detections/playbooks/`.
+- **GitHub PR integration** — detection proposals automatically create draft PRs against the tenant's detection repo when promoted. (`services/api/app/services/github.py`)
+- **BYOK per-tenant LLM credentials UI** — provider picker (OpenAI, Azure OpenAI, Anthropic, Ollama), API-key input, model selector, temperature slider, and connection test. (`SettingsView.tsx`, `llm_credentials.py`)
+- **WCAG AA accessibility** — axe-core CI gate covers 5 views + 3 modals; sidebar landmark roles, ARIA labels, focus trapping, skip-nav link, colour-contrast fixes. (`apps/web/src/test/a11y.test.tsx`)
+- **Light theme persisted in user profile** — stored in `localStorage`, synced to `PATCH /api/v1/users/me/preferences`. (`ThemeProvider.tsx`)
+- **Saved views + drag-drop dashboard widgets** — widgets can be dragged, dropped, resized, pinned, removed; layout serialised to `POST /api/v1/saved-views`. (`DashboardView.tsx`, `saved_views.py`)
+- **Threat actor attribution engine v0** — weighted IOC / TTP / Tool / Target scoring against three seed actor profiles (APT28, APT29, Lazarus). `POST /actors/attribute`. (`services/threatintel/app/actors/attribution.py`)
+- **Air-gap / Ollama local-LLM mode** — `docker-compose.airgap.yml` override; disables external feed pullers, enables Ollama sidecar; step-by-step deployment guide in docs. (`docker-compose.airgap.yml`, `apps/docs/docs/operations/air-gapped.md`)
+- **MSSP console improvements** — `GET /mssp/tenants` aggregation: per-child alert counts, open cases, SLA breach rate, last-seen connector heartbeat; `parent_tenant_id` + `mssp_role` added to `Tenant` model.
+- **Team analytics view** — analyst leaderboard, MTTR per analyst, cases closed per shift, FP rate trend. (`TeamAnalyticsView.tsx`)
+- **Air-gap LLM status endpoint** — reports whether air-gap mode is active and which Ollama models are available; drives the settings UI model picker. (`llm_status.py`)
Everything ships under MIT. Fork it, self-host it, audit it, extend it.
diff --git a/ROADMAP.md b/ROADMAP.md
index 58d4e4134..abf107526 100644
--- a/ROADMAP.md
+++ b/ROADMAP.md
@@ -168,15 +168,41 @@ SOAR platforms drove this release.
---
-## v7.0 — Planned
+## v7.0 — Shipped ✅ (2026-05-10)
+
+All items below were shipped as part of the v1.0 buyer-value plan.
+Implemented and reviewed by Beenu Arora <[email protected]>.
+
+- [x] WCAG AA full accessibility pass (axe-core CI gate — `apps/web/src/test/a11y.test.tsx`)
+- [x] Light theme persisted in user profile (`ThemeProvider.tsx` + `PATCH /api/v1/users/me/preferences`)
+- [x] Saved views and custom drag-drop dashboard widgets per analyst (`saved_views.py` + `DashboardView.tsx`)
+- [x] AI-generated weekly executive digest — auto-emailed PDF (`digest_pdf.py` + `weekly_digest_task.py`)
+- [x] Slack native bot for alert triage without opening the UI (`services/slack-bot/` — 61 tests)
+- [x] Threat actor attribution engine v0 (`services/threatintel/app/actors/attribution.py`)
+- [x] Air-gap / Ollama local-LLM mode (`docker-compose.airgap.yml` + `apps/docs/docs/operations/air-gapped.md`)
+- [x] BYOK per-tenant LLM credentials UI + API (`llm_credentials.py` + `SettingsView.tsx`)
+- [x] MSSP console — per-child-tenant KPI aggregation, SLA posture, parent_tenant_id hierarchy
+- [x] Team analytics view — analyst MTTR, leaderboard, shift workload (`TeamAnalyticsView.tsx`)
+- [x] Case auto-summary + PDF export (`case_summary.py` + `case_summary_html.py`)
+- [x] Investigation timeline (replayable) (`InvestigationTimeline.tsx`)
+- [x] Playbook gallery with 12 curated packs + GitHub PR integration for detection proposals
+- [ ] Mobile responder console (React Native) — triage and acknowledge from phone _(deferred to v8.0)_
+- [ ] Plugin publishing marketplace v3 (commercial plugins, revenue sharing) _(deferred to v8.0)_
+
+---
+
+## v8.0 — Planned
- Mobile responder console (React Native) — triage and acknowledge from phone
-- WCAG AA full accessibility pass
-- Light theme + brand-configurable white-label mode
-- Saved views and custom dashboard widgets per analyst
-- AI-generated weekly executive digest (auto-emailed PDF)
-- Slack / Teams native bot for alert triage without opening the UI
- Plugin publishing marketplace v3 (commercial plugins, revenue sharing)
+- MSSP RBAC enforcement on `/api/v1/actors/*` endpoints (threat attribution)
+- Automated IOC sharing to community MISP instances via STIX/TAXII push
+- NL→query: "show me failed logins from new ASNs last 24h" → ES|QL / KQL
+- AI-generated threat intelligence briefings from public feeds
+- Embedded red-team scoring (ATT&CK coverage %) as a live dashboard widget
+- SLA breach predictor (ML model on historical MTTR data)
+- Incident cost estimator (breach impact calculator)
+- SOC-in-a-box one-click cloud deploy (Terraform module for AWS / GCP)
---
diff --git a/apps/docs/docs/intro.md b/apps/docs/docs/intro.md
index 307040370..25270f548 100644
--- a/apps/docs/docs/intro.md
+++ b/apps/docs/docs/intro.md
@@ -41,7 +41,13 @@ public, reproducible eval harness on every PR targeting `main` / `develop`.
- **Hunt-as-Code** — YAML hunt definitions in `hunts/` with hypothesis, indicator matching (equals/in/regex/gte/lte/exists/contains_any/iendswith), and APScheduler-driven continuous execution. Results flow to Postgres via the Hunt API.
- **Risk-Based Alerting (RBA)** — alerts contribute time-decayed risk points to entities (user, host, IP, domain). When an entity's score crosses a configurable threshold, AiSOC promotes it to an incident with contributing alerts attached.
- **Federated search** — translate a single query into SPL, KQL, and ES|QL and fan out to connected SIEMs. Results are merged and deduplicated.
+- **Slack ChatOps bot** — dedicated `aisoc` slash commands (`/aisoc triage`, `/aisoc approve`, `/aisoc status`, `/aisoc summary`) with interactive approval buttons and HMAC-signed callbacks. Human-in-the-loop triage without opening the console. (`services/slack-bot/`)
- **ChatOps verification** — Slack/Teams interactive prompts with HMAC-signed callback choices (acknowledge / deny / escalate) for human-in-the-loop response actions.
+- **AI executive digest PDF** — branded A4 PDF with KPI tiles, alert-volume chart, top-rule table, and remediation summary, auto-emailed Monday 06:00 UTC via APScheduler. (`services/api/app/services/digest_pdf.py`)
+- **Threat actor attribution engine** — per-alert actor attribution with MITRE Group mapping, confidence scoring, Diamond Model labeling, and campaign clustering. (`services/threatintel/app/actors/attribution.py`)
+- **Air-gap / local-LLM mode** — run AiSOC with zero outbound HTTP using Ollama, LiteLLM, or vLLM via a single env-var toggle and a Docker Compose overlay. ([Air-gapped mode](./operations/air-gapped))
+- **BYOK — Bring Your Own LLM key** — per-tenant LLM credential management (API key, base URL, model) stored encrypted in `CredentialVault`; configurable from the Settings UI. ([Credentials](./operations/credentials))
+- **Saved views & drag-drop dashboard widgets** — analysts save custom alert filter presets and rearrange dashboard widgets; persisted per-user in Postgres.
- **Playbook engine** — 50+ community SOAR playbooks with explicit decision trees and human-approval gates on destructive actions.
- **Threat intelligence** — TAXII 2.1, MISP, OTX, CISA KEV with triple storage (search, vector, graph).
- **Governance** — SAML 2.0 + OIDC SSO, multi-tenant RLS, granular RBAC, immutable audit log.