Skip to content

Make Codex and Claude Code subscriptions go further: a local model router that gives routine agent work to economical models and keeps your strongest model for the hard parts.

feat(website): refresh homepage and illustrate Jev routing (#8)

澄潭e02d103a14cf
diff --git a/README.md b/README.md
index 74bb52ee..19205817 100644
--- a/README.md
+++ b/README.md
@@ -1,59 +1,173 @@
-# HiRoute
+<p align="right"><a href="README.zh-CN.md">简体中文</a></p>
-[简体中文](README.zh-CN.md)
+<p align="center">
+ <a href="https://hiroute.ai/">
+ <img src="apps/website/public/brand/github-social-preview.png" alt="HiRoute — intelligent routing for long-horizon agents" width="100%">
+ </a>
+</p>
-HiRoute is a local model-routing and agent-coordination engine for long-running tasks. It
-keeps model sources, routing plans, agent connections and execution evidence in one place,
-with a Desktop application, a headless CLI and a local daemon.
+<p align="center">
+ <strong>Route agents by task. Route models by stage.</strong><br>
+ A local-first routing and coordination engine for long-running agent work.
+</p>
-During a long task, HiRoute can reconsider the model at Agent-turn and context-hold
-boundaries. These are natural prefix changes, so routing decisions preserve the stable
-prefix within each stage and remain friendly to provider KV caches. The next decision can
-combine the new request with the previous stage's observed competence.
+<p align="center">
+ <a href="https://hiroute.ai/">Website</a> ·
+ <a href="https://hiroute.ai/download/">Download</a> ·
+ <a href="https://hiroute.ai/en/docs/">Documentation</a> ·
+ <a href="https://hiroute.ai/en/docs/decision-api/">Decision API</a> ·
+ <a href="CONTRIBUTING.md">Contributing</a>
+</p>
-## Capabilities
+<p align="center">
+ <a href="https://github.com/higress-group/HiRoute/releases/latest"><img alt="Latest release" src="https://img.shields.io/github/v/release/higress-group/HiRoute?display_name=tag&amp;sort=semver&amp;style=flat-square"></a>
+ <a href="LICENSE"><img alt="Apache-2.0 license" src="https://img.shields.io/badge/license-Apache--2.0-5A6FE8?style=flat-square"></a>
+ <a href="https://hiroute.ai/download/"><img alt="macOS and Linux" src="https://img.shields.io/badge/platform-macOS%20%7C%20Linux-171A21?style=flat-square"></a>
+</p>
-- Connect supported local subscriptions, registered APIs, and compatible custom APIs.
-- Publish fixed-model, smart-saving, free-first, and ordered fallback routing plans.
-- Configure model access and collaboration plans for supported Agent clients.
-- Run and manage Worker tasks from the CLI, with task state and results visible in Desktop.
-- Inspect local sessions, request execution, known usage and cost, and delete selected data.
-- Extend model selection through the public Decision API. The official Jev decider is a
- deployable reference implementation.
+HiRoute is a local control and execution layer for long-horizon agents. It brings model
+sources, reusable routing plans, agent connections, bounded failover, and execution evidence
+into one system, available through a Desktop application or a headless CLI and daemon.
-HiRoute is currently an MVP. The download page describes the platforms and architectures
-validated for each release; unsupported or untested environments are not implied by source
-availability.
+It is not a proxy that swaps models on every tool call. HiRoute keeps a selected branch stable
+through an execution stage, then chooses again at a natural boundary—such as new user input or
+the context rebuild that follows compaction. This preserves reusable prefixes within a stage
+and makes model switching friendly to provider KV caches.
+
+## Why HiRoute
+
+| | |
+| --- | --- |
+| **Route long tasks by stage** | Use the model that fits the work ahead instead of committing an entire task to one model. |
+| **Protect quality while saving** | Use observed competence to block risky downgrades, while task complexity creates opportunities to use economical models. |
+| **Keep handoffs bounded** | Enforce capability requirements, ordered candidates, and explicit exhaustion instead of silently changing behavior. |
+| **Learn from execution** | Retain the selected branch, actual model, tool outcomes, accepted output, and optional stage competence for later decisions. |
+
+HiRoute currently provides fixed-model, smart-saving, free-first, and ordered-fallback routing;
+agent work plans and Worker delegation; local session, usage, cost, and competence views; and a
+public Decision API for custom selection strategies.
## Get started
-- [Download HiRoute](https://hiroute.ai/download/)
-- [Documentation](https://hiroute.ai/docs/)
-- [macOS self-signed installation](docs/macos-installation.md)
-- [Linux headless and CLI](docs/standalone-cli.md)
-- [Model-selection decisions and extensions](decision-extensions/README.md)
-- [Decision API OpenAPI document](decision-extensions/api/decision.openapi.json)
+### macOS Desktop
+
+Download the current DMG from [hiroute.ai](https://hiroute.ai/download/), drag HiRoute into
+Applications, and follow the [self-signed package instructions](docs/macos-installation.md) if
+macOS asks for manual approval. The download page is the authority for validated operating
+systems and architectures.
+
+### Linux headless
+
+Install the current-user CLI and daemon without `sudo`:
+
+```sh
+curl -fsSL https://hiroute.ai/install.sh | sh
+hiroute service start --output json
+hiroute system status --output json
+```
+
+The installer verifies the published package manifest and checksums and does not start a
+service automatically. See the [Linux installation guide](https://hiroute.ai/en/docs/install-linux/)
+and [CLI reference](docs/standalone-cli.md) for service lifecycle and machine-readable commands.
+
+### Configure your first route
+
+1. Connect a supported subscription, registered API, or compatible custom API.
+2. Create and publish a routing plan for the task and cost profile you want.
+3. Connect an Agent client, or expose a Worker plan for delegated tasks.
+4. Inspect sessions and runtime performance to understand the selected branch and actual model.
+
+Continue with [model routing](https://hiroute.ai/en/docs/model-routing/),
+[task routing](https://hiroute.ai/en/docs/task-routing/), or the
+[headless CLI guide](https://hiroute.ai/en/docs/cli/).
+
+## See routing performance
+
+<p align="center">
+ <img src="decision-extensions/assets/quality-en.png" alt="HiRoute runtime performance showing rated, unrated, and partial model stages" width="100%">
+</p>
+
+The performance view follows the models selected by the active routing-plan revision; users do
+not need to enter internal model IDs. The screenshot uses realistic synthetic data rendered by
+real product components. It demonstrates the UI states and is not a model benchmark.
+
+## How routing works
+
+![Jev decision flow: one request selects the next branch and optionally assesses the prior stage](decision-extensions/assets/jev-decision-en.svg)
-For development, the Rust version is pinned by `rust-toolchain.toml`. The Desktop UI uses
-Node.js/npm and Tauri. See [CONTRIBUTING.md](CONTRIBUTING.md), the
-[Desktop development entry](apps/desktop/README.md), and the
-[website development and release guide](apps/website/README.md).
+A routing execution round begins with one branch decision. Ordinary tool continuations keep
+that decision while the current context remains reusable. HiRoute decides again when a new user
+request changes the work, or when a long session compacts and rebuilds its context. The routing
+engine determines whether a decision can be inherited; clients do not need to emit a separate
+compaction event. A provider or model failure can still use the plan's bounded fallback inside
+the round; fallback is distinct from a new classification decision.
-## Repository layout
+At a decision boundary, a service may do two related jobs in one response:
+
+- choose one of the branch IDs allowed by HiRoute for the next round;
+- optionally assess how competently the previous model handled its execution stage.
+
+The official [TypeSafe Jev extension](decision-extensions/extensions/jev-decider/README.md)
+implements this contract with one OpenRouter request. Its Rules policy uses task complexity
+and the optional assessment together: choose economy only when the simple-task probability
+meets the threshold and no valid current score falls below the competence floor. Otherwise,
+choose primary. A missing score leaves the decision to complexity alone.
+
+The principle is:
+
+> **Competence blocks risky cost-cutting; complexity creates opportunities to save.**
+
+Jev is optional. You can use built-in rules or implement the same general multi-branch contract
+with an LLM or your own policy. Start with the [decision mechanism](decision-extensions/README.md),
+the [API guide](decision-extensions/api/README.md), or the canonical
+[OpenAPI document](decision-extensions/api/decision.openapi.json).
+
+## For technical contributors
+
+HiRoute is a Rust workspace with a Tauri Desktop client and an Astro website. The Rust toolchain
+is pinned in `rust-toolchain.toml`; Desktop uses Node.js 24 and the website uses Node.js 22.
+
+```sh
+git clone https://github.com/higress-group/HiRoute.git
+cd HiRoute
+
+cargo fmt --check
+cargo test --locked --workspace --exclude hiroute-desktop --all-features
+```
+
+For frontend work, run the checks in the affected application:
+
+```sh
+cd apps/desktop # or apps/website
+npm ci --ignore-scripts
+npm test
+npm run build
+```
+
+See [CONTRIBUTING.md](CONTRIBUTING.md) for native dependencies, Clippy, validation scope, and
+language conventions. Native Desktop behavior and real-provider integration require their
+respective environments; a source build alone is not product acceptance.
+
+### Repository map
| Path | Responsibility |
| --- | --- |
| `apps/desktop` | Desktop UI and Tauri host |
-| `apps/website` | Bilingual `hiroute.ai` site, downloads and OSS publication |
-| `crates/cli`, `crates/daemon` | Command-line client and local service |
-| `crates/client-core`, `crates/application` | Shared client and application orchestration |
-| `crates/gateway-core`, `crates/gateway` | Protocol adapters, request execution and model fallback |
-| `crates/observation`, `crates/local-storage` | Local execution evidence, query and persistence |
-| `contracts`, `assets` | Runtime contracts, model metadata and product resources |
-| `decision-extensions` | Decision mechanism, OpenAPI contract and official extensions |
-| `tools`, `e2e` | Validation utilities and executable product scenarios |
-
-## License and security
-
-HiRoute is licensed under [Apache License 2.0](LICENSE). Report security issues and crash
-reports through the process in [SECURITY.md](SECURITY.md).
+| `crates/cli`, `crates/daemon` | Headless client and local role-all service |
+| `crates/application`, `crates/client-core` | Application workflows and shared client behavior |
+| `crates/gateway`, `crates/gateway-core` | Agent-facing protocols, routing, execution, and fallback |
+| `crates/observation`, `crates/local-storage` | Local execution evidence, queries, and persistence |
+| `decision-extensions` | Decision mechanism, OpenAPI contract, screenshots, and official Jev service |
+| `contracts`, `assets` | Current runtime contracts, model metadata, and product resources |
+| `apps/website` | Bilingual `hiroute.ai` site, downloads, and release publication |
+| `e2e`, `tools`, `scripts` | Product scenarios, validation tools, packaging, and automation |
+
+## Project status
+
+HiRoute is an MVP. Platform and architecture claims are tied to the evidence for each release;
+source availability does not imply that every environment has been validated. Current Agent
+integrations and planned directions remain separate on the [website](https://hiroute.ai/).
+
+HiRoute is licensed under the [Apache License 2.0](LICENSE). Use
+[GitHub Issues](https://github.com/higress-group/HiRoute/issues) for bugs and feature requests.
+Report security issues and crash reports through the process in [SECURITY.md](SECURITY.md).
diff --git a/README.zh-CN.md b/README.zh-CN.md
index 7447c997..37bd7064 100644
--- a/README.zh-CN.md
+++ b/README.zh-CN.md
@@ -1,54 +1,162 @@
-# HiRoute
+<p align="right"><a href="README.md">English</a></p>
-[English](README.md)
+<p align="center">
+ <a href="https://hiroute.ai/">
+ <img src="apps/website/public/brand/github-social-preview.png" alt="HiRoute——面向长程 Agent 的智能路由" width="100%">
+ </a>
+</p>
-HiRoute 是面向长程任务、本地运行的模型路由与 Agent 协作引擎。它将模型来源、路由计划、Agent
-接入和执行证据集中管理,提供 Desktop、headless CLI 与本地后台服务。
+<p align="center">
+ <strong>按任务选择 Agent,按阶段选择模型。</strong><br>
+ 面向长程 Agent 工作、本地优先的智能路由与协作引擎。
+</p>
-在长程任务中,HiRoute 可以在 Agent turn 和 ContextHold 边界重新选择模型。这些位置本身会改变
-稳定前缀,因此每个阶段内仍保持前缀稳定,对模型供应商的 KV cache 友好。下一次决策可以结合新的
-用户请求与上一阶段的胜任度表现。
+<p align="center">
+ <a href="https://hiroute.ai/">官网</a> ·
+ <a href="https://hiroute.ai/download/">下载</a> ·
+ <a href="https://hiroute.ai/docs/">使用文档</a> ·
+ <a href="https://hiroute.ai/docs/decision-api/">决策 API</a> ·
+ <a href="CONTRIBUTING.md">参与贡献</a>
+</p>
-## 当前能力
+<p align="center">
+ <a href="https://github.com/higress-group/HiRoute/releases/latest"><img alt="最新版本" src="https://img.shields.io/github/v/release/higress-group/HiRoute?display_name=tag&amp;sort=semver&amp;style=flat-square"></a>
+ <a href="LICENSE"><img alt="Apache-2.0 许可证" src="https://img.shields.io/badge/license-Apache--2.0-5A6FE8?style=flat-square"></a>
+ <a href="https://hiroute.ai/download/"><img alt="支持 macOS 和 Linux" src="https://img.shields.io/badge/platform-macOS%20%7C%20Linux-171A21?style=flat-square"></a>
+</p>
-- 接入受支持的本机订阅、注册 API 和兼容的自定义 API。
-- 发布固定模型、智能省钱、免费优先和有序故障接力路由计划。
-- 为受支持的 Agent 客户端配置模型接入和协作计划。
-- 通过 CLI 运行和管理 Worker 任务,并在 Desktop 查看任务状态与结果。
-- 查看本地会话、请求执行、已知用量与成本,并删除选定数据。
-- 通过公开 Decision API 扩展模型选择;官方 Jev 决策器提供可部署的参考实现。
+HiRoute 是长程 Agent 的本地控制与执行层。它把模型来源、可复用路由计划、Agent 接入、有边界的
+故障接力和执行证据放在同一套系统里,同时提供 Desktop 应用以及无界面 CLI 和后台服务。
-HiRoute 当前处于 MVP 阶段。下载页会说明各版本已验证的平台和架构;源码存在不代表未验证环境已经
-完成产品验收。
+HiRoute 不是在每次工具调用时随意换模型的代理。一次执行阶段内会保持已选分支稳定,仅在新的用户
+输入或上下文压缩后的自然重建时机重新选择。这样既能按阶段使用合适的模型,也能让阶段内前缀持续
+复用,对模型供应商的 KV cache 友好。
-## 开始使用
+## 为什么使用 HiRoute
-- [下载 HiRoute](https://hiroute.ai/download/)
-- [使用文档](https://hiroute.ai/docs/)
-- [macOS 自签名安装](docs/macos-installation.zh-CN.md)
-- [Linux headless 与 CLI](docs/standalone-cli.zh-CN.md)
-- [模型选择决策机制与扩展](decision-extensions/README.zh-CN.md)
-- [Decision API OpenAPI 文档](decision-extensions/api/decision.openapi.json)
+| | |
+| --- | --- |
+| **按阶段路由长任务** | 不必让一个模型包办整段任务,而是根据后续工作选择合适的模型。 |
+| **在降本时守住质量** | 胜任度用于阻止冒险降本,任务复杂度用于寻找采用经济模型的机会。 |
+| **让接力有明确边界** | 校验能力要求、按顺序尝试候选,候选耗尽时明确停止,不静默改变行为。 |
+| **让执行成为下一次决策的证据** | 保留选择分支、实际模型、工具结果、接受的输出和可选阶段胜任度。 |
+
+HiRoute 当前提供固定模型、智能省钱、免费优先和有序故障接力路由;Agent 工作计划与 Worker
+委派;本地会话、用量、成本和胜任度查看;以及用于扩展选择策略的公开决策 API。
+
+## 快速开始
+
+### macOS Desktop
+
+从 [hiroute.ai](https://hiroute.ai/download/) 下载当前 DMG,拖入“应用程序”目录。如果 macOS
+要求手动批准,请按[自签名安装说明](docs/macos-installation.zh-CN.md)操作。每个版本已验证的系统和
+架构以下载页为准。
+
+### Linux 无界面版
+
+无需 `sudo`,安装到当前用户目录:
+
+```sh
+curl -fsSL https://hiroute.ai/install.sh | sh
+hiroute service start --output json
+hiroute system status --output json
+```
+
+安装器会校验已发布的 package manifest 与文件摘要,并且不会自动启动服务。服务生命周期和
+机器可读命令见 [Linux 安装指南](https://hiroute.ai/docs/install-linux/)与
+[CLI 参考](docs/standalone-cli.zh-CN.md)。
+
+### 配置第一条路由
+
+1. 接入受支持的本机订阅、注册 API 或兼容的自定义 API。
+2. 根据任务与成本目标创建并发布路由计划。
+3. 接入 Agent 客户端,或开放 Worker 计划供主 Agent 委派任务。
+4. 在会话和运行表现中查看选择分支、实际模型和执行结果。
+
+继续阅读[智能模型路由](https://hiroute.ai/docs/model-routing/)、
+[智能任务路由](https://hiroute.ai/docs/task-routing/)或
+[无界面 CLI 指南](https://hiroute.ai/docs/cli/)。
+
+## 查看真实产品中的运行表现
+
+<p align="center">
+ <img src="decision-extensions/assets/quality-zh-CN.png" alt="HiRoute 运行表现,展示已评分、未评分和部分证据阶段" width="100%">
+</p>
+
+运行表现会自动展示当前路由计划生效版本中的模型,用户无需填写内部模型 ID。截图由真实产品组件
+配合贴近实际的模拟数据生成,用于说明界面状态,不代表模型能力评测。
+
+## 路由机制
-参与开发时,Rust 版本由 `rust-toolchain.toml` 固定,Desktop UI 使用 Node.js/npm 与 Tauri。
-请阅读 [贡献指南](CONTRIBUTING.md)、[Desktop 开发入口](apps/desktop/README.zh-CN.md)和
-[官网开发及发布指南](apps/website/README.md)。
+![Jev 决策机制:一次请求选择下一分支,并可评价此前阶段](decision-extensions/assets/jev-decision-zh-CN.svg)
-## 代码结构
+一个路由执行轮次从一次分支选择开始。当前上下文可以持续复用时,普通工具续轮会保持该选择;新的
+用户请求改变工作目标,或长会话发生压缩并重建上下文时,HiRoute 会再次决策。是否能继承已有决策
+由路由引擎判断,客户端无需额外上报压缩事件。供应商或模型失败仍可在本轮内按计划执行有边界的
+故障接力,它与重新分类是两件事。
+
+在一个决策边界,同一次服务响应可以完成两个相关工作:
+
+- 从 HiRoute 允许的 branch ID 中选择下一轮执行分支;
+- 可选地评价上一模型在执行阶段中的胜任程度。
+
+官方 [TypeSafe Jev 扩展](decision-extensions/extensions/jev-decider/README.zh-CN.md)通过一次
+OpenRouter 请求实现这套协议。其 Rules 策略结合任务复杂度与可选胜任度:任务简单的概率达到门槛,
+且本次合法评分未低于胜任度下限时,选择经济分支;否则选择主力分支。缺少评分时,仅依据复杂度判断。
+
+这套机制的原则是:
+
+> **胜任度只负责阻止冒险降本,复杂度负责提供降本机会。**
+
+Jev 是可选扩展。你也可以使用内置规则,或通过 LLM、自有模型和规则实现同一套通用多分支协议。
+可从[决策机制](decision-extensions/README.zh-CN.md)、
+[API 说明](decision-extensions/api/README.zh-CN.md)或规范化
+[OpenAPI 文档](decision-extensions/api/decision.openapi.json)开始。
+
+## 面向技术开发者
+
+HiRoute 是 Rust workspace,Desktop 使用 Tauri,官网使用 Astro。Rust 工具链由
+`rust-toolchain.toml` 固定;Desktop 使用 Node.js 24,官网使用 Node.js 22。
+
+```sh
+git clone https://github.com/higress-group/HiRoute.git
+cd HiRoute
+
+cargo fmt --check
+cargo test --locked --workspace --exclude hiroute-desktop --all-features
+```
+
+前端改动请在受影响的应用目录运行对应检查:
+
+```sh
+cd apps/desktop # 或 apps/website
+npm ci --ignore-scripts
+npm test
+npm run build
+```
+
+原生依赖、Clippy、验证范围和语言约定见[贡献指南](CONTRIBUTING.md)。macOS 原生 Desktop 行为和
+真实供应商集成需要相应环境,源码构建成功本身不等于产品验收通过。
+
+### 代码结构
| 目录 | 职责 |
| --- | --- |
-| `apps/desktop` | Desktop 页面与 Tauri 宿主 |
-| `apps/website` | `hiroute.ai` 双语官网、下载与 OSS 发布 |
-| `crates/cli`、`crates/daemon` | 命令行客户端和本地服务 |
-| `crates/client-core`、`crates/application` | 共享客户端与业务编排 |
-| `crates/gateway-core`、`crates/gateway` | 协议适配、请求执行与模型故障接力 |
+| `apps/desktop` | Desktop 界面与 Tauri 宿主 |
+| `crates/cli`、`crates/daemon` | 无界面客户端与本地 role-all 服务 |
+| `crates/application`、`crates/client-core` | 应用工作流和共享客户端行为 |
+| `crates/gateway`、`crates/gateway-core` | Agent 协议入口、路由、执行与故障接力 |
| `crates/observation`、`crates/local-storage` | 本地执行证据、查询与持久化 |
-| `contracts`、`assets` | 运行合同、模型资料与产品资源 |
-| `decision-extensions` | 决策机制、OpenAPI 合同与官方扩展 |
-| `tools`、`e2e` | 验证工具与可执行产品场景 |
+| `decision-extensions` | 决策机制、OpenAPI、产品截图和官方 Jev 服务 |
+| `contracts`、`assets` | 当前运行合同、模型元数据和产品资源 |
+| `apps/website` | `hiroute.ai` 双语官网、下载和版本发布 |
+| `e2e`、`tools`、`scripts` | 产品场景、验证工具、打包和自动化 |
+
+## 项目状态
-## 许可与安全
+HiRoute 当前处于 MVP 阶段。平台和架构声明与每个版本的验证证据绑定;源码存在不代表所有环境均已
+验收。当前可用的 Agent 集成与后续方向会在[官网](https://hiroute.ai/)中明确区分。
-HiRoute 使用 [Apache License 2.0](LICENSE)。安全问题和 crash report 请按
-[SECURITY.md](SECURITY.md) 提交。
+HiRoute 使用 [Apache License 2.0](LICENSE)。缺陷和功能建议请提交到
+[GitHub Issues](https://github.com/higress-group/HiRoute/issues)。安全问题和 crash report 请按
+[SECURITY.md](SECURITY.md)中的流程提交。
diff --git a/apps/website/content/guides/install-linux.en.md b/apps/website/content/guides/install-linux.en.md
index 51d0f5d1..a3b3fc7a 100644
--- a/apps/website/content/guides/install-linux.en.md
+++ b/apps/website/content/guides/install-linux.en.md
@@ -26,7 +26,7 @@ The entry points are installed under `~/.local/bin`. If your shell cannot find `
export PATH="$HOME/.local/bin:$PATH"
```
-The website script selects only the latest stable Linux package in the public release manifest. The existing installer verifies the target architecture, archive SHA256, closed file inventory, and every file digest; the website does not create a second installation implementation. Desktop and standalone must not manage the service in the same `HOME`.
+The installer downloads the latest stable Linux release and checks its architecture and file integrity before installation. Desktop and standalone must not manage the service in the same `HOME`.
## 2. Start the service explicitly
diff --git a/apps/website/content/guides/install-linux.zh.md b/apps/website/content/guides/install-linux.zh.md
index 49ca0ce4..f09e2c61 100644
--- a/apps/website/content/guides/install-linux.zh.md
+++ b/apps/website/content/guides/install-linux.zh.md
@@ -26,7 +26,7 @@ sh install.sh
export PATH="$HOME/.local/bin:$PATH"
```
-安装脚本只选择官网发布清单中的最新稳定 Linux 包。现有安装器会核对目标架构、归档 SHA256、闭合文件清单和逐文件摘要;官网脚本不复制第二套安装逻辑。桌面应用与独立服务版(standalone)不应在同一个 `HOME` 中同时管理服务。
+安装脚本会下载最新稳定的 Linux 版本,并在安装前检查架构与文件完整性。桌面应用与独立服务版(standalone)不应在同一个 `HOME` 中同时管理服务。
## 2. 显式启动服务
diff --git a/apps/website/content/guides/model-routing.en.md b/apps/website/content/guides/model-routing.en.md
index 1beb970d..1d280f58 100644
--- a/apps/website/content/guides/model-routing.en.md
+++ b/apps/website/content/guides/model-routing.en.md
@@ -31,7 +31,7 @@ The service receives accepted assistant text plus ordered tool names and coarse
## Understand decision boundaries
-New user input creates a decision opportunity. During autonomous work, HiRoute also chooses again—and may assess the previous stage—whenever ContextHold says the current context boundary permits a switch. The client does not need to report a separate “compaction happened” signal. Consecutive tool continuations stay within the current stage.
+New user input creates a decision opportunity. During autonomous work, compaction and the context rebuild that follows provide another natural opportunity to choose again and assess the previous stage. HiRoute's routing engine determines whether the existing decision can still be inherited; the client does not need to detect or report compaction separately. Consecutive tool continuations stay within the current stage.
A new decision does not force a model change. HiRoute can keep the existing branch when it still fits. Different models cannot share a KV cache, but deciding at a boundary where context already needs to be rebuilt avoids breaking a stable prefix merely to classify an ordinary tool continuation.
diff --git a/apps/website/content/guides/model-routing.zh.md b/apps/website/content/guides/model-routing.zh.md
index b7660ed6..2c989d29 100644
--- a/apps/website/content/guides/model-routing.zh.md
+++ b/apps/website/content/guides/model-routing.zh.md
@@ -31,7 +31,7 @@
## 理解切换边界
-新用户输入会产生新的决策机会。Agent 自主工作期间,只要 ContextHold 判断当前上下文边界允许切换,HiRoute 也会重新选择分支并可评价上一阶段;不需要客户端另报“发生了压缩”。连续追加的工具调用仍留在当前阶段。
+新用户输入会产生新的决策机会。Agent 自主工作期间,上下文压缩及随后的重建也会成为重新选择分支、评价上一阶段的自然机会。是否能继承已有决策由 HiRoute 路由引擎判断,客户端无需额外识别或上报压缩事件。连续追加的工具调用仍留在当前阶段。
重新决策不等于一定换模。如果原分支仍适合后续工作,HiRoute 可以继续使用它。不同模型不能共享 KV cache,但把选择放在本来就要重建上下文的边界,可以避免为了分类在普通工具续轮中主动打断稳定前缀。
diff --git a/apps/website/public/brand/display-wordmark.svg b/apps/website/public/brand/display-wordmark.svg
new file mode 100644
index 00000000..3eea020e
--- /dev/null
+++ b/apps/website/public/brand/display-wordmark.svg
@@ -0,0 +1,21 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 590 198" fill="none" role="img" aria-labelledby="title description">
+ <title id="title">HiRoute display wordmark</title>
+ <desc id="description">HiRoute geometric wordmark with a blue-purple routing arc.</desc>
+ <defs>
+ <linearGradient id="route" x1="223" y1="56" x2="486" y2="15" gradientUnits="userSpaceOnUse">
+ <stop stop-color="#5A6FE8"/>
+ <stop offset=".58" stop-color="#9AA7FF"/>
+ <stop offset="1" stop-color="#8B73E4"/>
+ </linearGradient>
+ </defs>
+ <path d="M223 61C288 20 398 -7 465 6C499 13 507 39 489 62C486 66 482 69 478 72L474 72C486 58 491 41 480 29C454 -3 330 15 223 61Z" fill="url(#route)"/>
+ <g fill="#171A21" fill-rule="evenodd" transform="translate(8 58)">
+ <path d="M0 16H18V57H69V16H87V120H69V75H18V120H0Z"/>
+ <path d="M103 19Q103 16 106 16H118Q121 16 121 19V31Q121 34 118 34H106Q103 34 103 31ZM103 43H121V120H103Z"/>
+ <path d="M141 16H182C210 16 224 29 224 51C224 67 215 78 201 82L229 120H206L180 84H160V120H141ZM160 33V67H181C196 67 205 62 205 50C205 39 197 33 181 33Z"/>
+ <path d="M279 41C303 41 320 58 320 82C320 106 303 122 279 122C255 122 238 106 238 82C238 58 255 41 279 41ZM279 58C265 58 256 67 256 82C256 97 265 106 279 106C293 106 302 97 302 82C302 67 293 58 279 58Z"/>
+ <path d="M335 43H353V87C353 100 359 106 370 106C382 106 390 97 390 83V43H408V120H391V110C385 118 376 122 365 122C346 122 335 110 335 90Z"/>
+ <path d="M436 22H454V43H477V59H454V93C454 102 458 106 466 106C470 106 474 105 478 103V119C473 121 467 122 461 122C444 122 436 113 436 96V59H424V43H436Z"/>
+ <path d="M570 85H509C510 98 519 106 533 106C543 106 551 103 558 97L569 108C559 118 548 122 533 122C507 122 491 106 491 82C491 58 507 41 531 41C554 41 571 56 571 80C571 82 571 83 570 85ZM509 71H553C551 61 544 56 532 56C521 56 513 62 509 71Z"/>
+ </g>
+</svg>
diff --git a/apps/website/public/brand/github-social-preview.png b/apps/website/public/brand/github-social-preview.png
new file mode 100644
index 00000000..ac729676
Binary files /dev/null and b/apps/website/public/brand/github-social-preview.png differ
diff --git a/apps/website/public/brand/github-social-preview.svg b/apps/website/public/brand/github-social-preview.svg
new file mode 100644
index 00000000..4fa204b2
--- /dev/null
+++ b/apps/website/public/brand/github-social-preview.svg
@@ -0,0 +1,66 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="1280" height="640" viewBox="0 0 1280 640" fill="none" role="img" aria-labelledby="title description">
+ <title id="title">HiRoute repository social preview</title>
+ <desc id="description">HiRoute display wordmark and a staged routing path on a deep graphite background.</desc>
+ <defs>
+ <linearGradient id="background" x1="82" y1="42" x2="1198" y2="598" gradientUnits="userSpaceOnUse">
+ <stop stop-color="#11141D"/>
+ <stop offset=".58" stop-color="#0B0D12"/>
+ <stop offset="1" stop-color="#10121A"/>
+ </linearGradient>
+ <linearGradient id="route" x1="427" y1="190" x2="864" y2="126" gradientUnits="userSpaceOnUse">
+ <stop stop-color="#5A6FE8"/>
+ <stop offset=".58" stop-color="#9AA7FF"/>
+ <stop offset="1" stop-color="#8B73E4"/>
+ </linearGradient>
+ <linearGradient id="trail" x1="390" y1="500" x2="890" y2="500" gradientUnits="userSpaceOnUse">
+ <stop stop-color="#5A6FE8" stop-opacity="0"/>
+ <stop offset=".14" stop-color="#697DF0"/>
+ <stop offset=".53" stop-color="#9AA7FF"/>
+ <stop offset=".86" stop-color="#8B73E4"/>
+ <stop offset="1" stop-color="#8B73E4" stop-opacity="0"/>
+ </linearGradient>
+ <radialGradient id="glow" cx="0" cy="0" r="1" gradientTransform="translate(640 265) rotate(90) scale(286 518)" gradientUnits="userSpaceOnUse">
+ <stop stop-color="#5A6FE8" stop-opacity=".14"/>
+ <stop offset="1" stop-color="#5A6FE8" stop-opacity="0"/>
+ </radialGradient>
+ <filter id="blur" x="191" y="-44" width="898" height="618" filterUnits="userSpaceOnUse" color-interpolation-filters="sRGB">
+ <feGaussianBlur stdDeviation="56"/>
+ </filter>
+ </defs>
+
+ <rect width="1280" height="640" fill="url(#background)"/>
+ <ellipse cx="640" cy="265" rx="337" ry="197" fill="url(#glow)" filter="url(#blur)"/>
+
+ <g opacity=".22" stroke="#7683AD" stroke-width="1">
+ <path d="M80 104H1200"/>
+ <path d="M80 536H1200"/>
+ <path d="M176 64V576"/>
+ <path d="M1104 64V576"/>
+ </g>
+
+ <!-- Complete display wordmark. Geometry is shared with brand/display-wordmark.svg. -->
+ <g transform="translate(345 126)">
+ <path d="M223 61C288 20 398 -7 465 6C499 13 507 39 489 62C486 66 482 69 478 72L474 72C486 58 491 41 480 29C454 -3 330 15 223 61Z" fill="url(#route)"/>
+ <g fill="#F3F5FA" fill-rule="evenodd" transform="translate(8 58)">
+ <path d="M0 16H18V57H69V16H87V120H69V75H18V120H0Z"/>
+ <path d="M103 19Q103 16 106 16H118Q121 16 121 19V31Q121 34 118 34H106Q103 34 103 31ZM103 43H121V120H103Z"/>
+ <path d="M141 16H182C210 16 224 29 224 51C224 67 215 78 201 82L229 120H206L180 84H160V120H141ZM160 33V67H181C196 67 205 62 205 50C205 39 197 33 181 33Z"/>
+ <path d="M279 41C303 41 320 58 320 82C320 106 303 122 279 122C255 122 238 106 238 82C238 58 255 41 279 41ZM279 58C265 58 256 67 256 82C256 97 265 106 279 106C293 106 302 97 302 82C302 67 293 58 279 58Z"/>
+ <path d="M335 43H353V87C353 100 359 106 370 106C382 106 390 97 390 83V43H408V120H391V110C385 118 376 122 365 122C346 122 335 110 335 90Z"/>
+ <path d="M436 22H454V43H477V59H454V93C454 102 458 106 466 106C470 106 474 105 478 103V119C473 121 467 122 461 122C444 122 436 113 436 96V59H424V43H436Z"/>
+ <path d="M570 85H509C510 98 519 106 533 106C543 106 551 103 558 97L569 108C559 118 548 122 533 122C507 122 491 106 491 82C491 58 507 41 531 41C554 41 571 56 571 80C571 82 571 83 570 85ZM509 71H553C551 61 544 56 532 56C521 56 513 62 509 71Z"/>
+ </g>
+ </g>
+
+ <text x="640" y="385" fill="#B9C1D4" font-family="Inter, -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif" font-size="27" font-weight="450" letter-spacing=".25" text-anchor="middle">Intelligent routing for long-horizon agents</text>
+
+ <g aria-hidden="true">
+ <path d="M390 500H890" stroke="url(#trail)" stroke-width="2"/>
+ <circle cx="445" cy="500" r="6" fill="#697DF0"/>
+ <circle cx="575" cy="500" r="6" fill="#7E8FF7"/>
+ <circle cx="705" cy="500" r="6" fill="#9AA7FF"/>
+ <circle cx="835" cy="500" r="6" fill="#8B73E4"/>
+ <circle cx="705" cy="500" r="15" stroke="#9AA7FF" stroke-opacity=".22"/>
+ </g>
+ <text x="640" y="554" fill="#727B91" font-family="ui-monospace, SFMono-Regular, Menlo, Consolas, monospace" font-size="16" letter-spacing="2" text-anchor="middle">CHOOSE AGENTS BY TASK · CHOOSE MODELS BY STAGE</text>
+</svg>
diff --git a/apps/website/public/scripts/home.js b/apps/website/public/scripts/home.js
index 2ec9be4a..5a4981a1 100644
--- a/apps/website/public/scripts/home.js
+++ b/apps/website/public/scripts/home.js
@@ -2,8 +2,8 @@
const boundaries = document.documentElement.lang === 'en' ? {
compaction: {
kicker: 'A decision point during autonomous work',
- title: 'A rebuilt context is a chance to choose again.',
- copy: 'When ContextHold permits a switch, HiRoute reassesses the work ahead; the client need not send a separate compaction signal.',
+ title: 'After context compaction, choose again.',
+ copy: 'Compaction rebuilds a long session around a new prefix. HiRoute uses that moment to assess the prior stage and the work ahead; the client does not need to detect or report compaction separately.',
assessment: 'Assess the work already performed by model A',
choice: 'Choose the primary branch; model B takes over'
},
@@ -17,8 +17,8 @@ const boundaries = document.documentElement.lang === 'en' ? {
} : {
compaction: {
kicker: '自主执行中的决策机会',
- title: '重建上下文时,可以重新选择。',
- copy: 'ContextHold 判断当前边界允许切换时,HiRoute 重新评估下一段工作;不要求客户端额外报告“发生了压缩”。',
+ title: '上下文压缩后,可以重新选择。',
+ copy: '长会话压缩并重建上下文时,可复用前缀本来就会变化。HiRoute 利用这个时机评估上一阶段和后续工作;客户端无需额外识别或上报压缩事件。',
assessment: '回看模型 A 已经完成的执行片段',
choice: '选择主力分支,让模型 B 接续'
},
diff --git a/apps/website/scripts/prepare-content.mjs b/apps/website/scripts/prepare-content.mjs
index af0261e5..e66bc39a 100644
--- a/apps/website/scripts/prepare-content.mjs
+++ b/apps/website/scripts/prepare-content.mjs
@@ -20,7 +20,7 @@ await fs.mkdir(assetsDirectory, { recursive: true });
await fs.mkdir(installDirectory, { recursive: true });
await fs.copyFile(path.join(source, 'api/decision.openapi.json'), path.join(apiDirectory, 'decision.openapi.json'));
for (const entry of await fs.readdir(path.join(source, 'assets'), { withFileTypes: true })) {
- if (entry.isFile() && entry.name.endsWith('.png')) {
+ if (entry.isFile() && /\.(png|svg)$/.test(entry.name)) {
await fs.copyFile(path.join(source, 'assets', entry.name), path.join(assetsDirectory, entry.name));
}
}
diff --git a/apps/website/src/components/ChangelogPage.astro b/apps/website/src/components/ChangelogPage.astro
index 16bbcbc0..4d37f626 100644
--- a/apps/website/src/components/ChangelogPage.astro
+++ b/apps/website/src/components/ChangelogPage.astro
@@ -11,6 +11,6 @@ const alternatePath = en ? '/changelog/' : '/en/changelog/';
<SiteLayout language={language} title={t('更新日志', 'Changelog') + ' · HiRoute'}
description={t('HiRoute 公开版本更新记录。', 'Public HiRoute release notes.')}
path={path} alternatePath={alternatePath}>
- <section class="page-hero wrap"><div class="eyebrow section-label">{t('更新日志', 'CHANGELOG')}</div><h1>{t('更新日志', 'Changelog')}</h1><p>{t('与下载页共用同一份发布清单,不单独维护容易漂移的版本列表。', 'Generated from the same release manifest as the download page—no separate version list to drift.')}</p></section>
+ <section class="page-hero wrap"><div class="eyebrow section-label">{t('更新日志', 'CHANGELOG')}</div><h1>{t('更新日志', 'Changelog')}</h1><p>{t('了解每个版本的新功能与改进。', 'Explore new features and improvements in each release.')}</p></section>
<section class="content-section wrap changelog-list">{releaseManifest.releases.length === 0 ? <div class="empty-release"><h2>{t('尚无公开版本', 'No public releases yet')}</h2><p>{t('首个通过公开发布门槛的版本会显示在这里。', 'The first release admitted by the public release gates will appear here.')}</p></div> : releaseManifest.releases.map(release => <article><span>{channelLabel(release.channel)}</span><h2>HiRoute {release.version}</h2><time datetime={release.published_at}>{new Intl.DateTimeFormat(en ? 'en' : 'zh-CN', { dateStyle: 'long' }).format(new Date(release.published_at))}</time><p>{release.notes[language]}</p><a href={en ? '/en/download/' : '/download/'}>{t('下载与摘要', 'Downloads and checksums')} →</a></article>)}</section>
</SiteLayout>
diff --git a/apps/website/src/components/DownloadPage.astro b/apps/website/src/components/DownloadPage.astro
index 174a1d20..a4869ef8 100644
--- a/apps/website/src/components/DownloadPage.astro
+++ b/apps/website/src/components/DownloadPage.astro
@@ -16,10 +16,10 @@ const linuxArchitectures = stableLinux.map(artifact => artifact.architecture).jo
<SiteLayout language={language} title={t('下载 HiRoute', 'Download HiRoute') + ' · HiRoute'}
description={t('在 macOS 使用桌面应用,或在 Linux 用一条命令安装 HiRoute 无界面版。', 'Use HiRoute Desktop on macOS or install HiRoute headless on Linux with one command.')}
path={path} alternatePath={alternatePath}>
- <section class="page-hero wrap"><div class="eyebrow section-label">{t('下载', 'DOWNLOAD')}</div><h1>{t('选择你的运行方式', 'Choose how to run HiRoute')}</h1><p>{t('macOS 提供可视化桌面应用;Linux 提供面向服务器和自动化环境的无界面命令行。发布文件都来自 hiroute.ai 的不可变 OSS 对象,并公开版本与 SHA256。', 'macOS provides the visual Desktop app. Linux provides a headless CLI for servers and automation. Every published file is an immutable hiroute.ai OSS object with a declared version and SHA256.')}</p></section>
+ <section class="page-hero wrap"><div class="eyebrow section-label">{t('下载', 'DOWNLOAD')}</div><h1>{t('选择你的运行方式', 'Choose how to run HiRoute')}</h1><p>{t('macOS 提供可视化桌面应用;Linux 提供面向服务器和自动化环境的无界面命令行。选择适合你的安装包,并可通过 SHA256 校验文件完整性。', 'macOS provides the visual Desktop app. Linux provides a headless CLI for servers and automation. Choose the package for your platform and verify its integrity with the provided SHA256 checksum.')}</p></section>
<section class="content-section wrap platform-install">
- <div class="platform-heading"><div><span class="release-channel">{t('Linux · 无界面版', 'Linux · headless')}</span><h2>{t('一条命令,按当前用户安装', 'One command, installed for the current user')}</h2><p>{t('安装器识别 x86_64 或 aarch64,下载稳定版的不可变清单和归档,验证内容后安装到 ~/.local;不使用 sudo,也不会自动启动服务。', 'The installer detects x86_64 or aarch64, downloads the stable immutable manifest and archive, verifies them, and installs under ~/.local. It uses no sudo and does not start the service automatically.')}</p></div><a class="text-link" href={en ? '/en/docs/install-linux/' : '/docs/install-linux/'}>{t('Linux 无界面版指南', 'Linux headless guide')} →</a></div>
+ <div class="platform-heading"><div><span class="release-channel">{t('Linux · 无界面版', 'Linux · headless')}</span><h2>{t('一条命令,按当前用户安装', 'One command, installed for the current user')}</h2><p>{t('安装器识别 x86_64 或 aarch64,下载最新稳定版,校验后安装到 ~/.local;不使用 sudo,也不会自动启动服务。', 'The installer detects x86_64 or aarch64, downloads the latest stable release, verifies it, and installs under ~/.local. It uses no sudo and does not start the service automatically.')}</p></div><a class="text-link" href={en ? '/en/docs/install-linux/' : '/docs/install-linux/'}>{t('Linux 无界面版指南', 'Linux headless guide')} →</a></div>
<div class:list={['install-command', { unavailable: stableLinux.length === 0 }]}>
<span>{stableLinux.length > 0 ? t(`稳定版 ${stable.version} · ${linuxArchitectures}`, `Stable ${stable.version} · ${linuxArchitectures}`) : t('首个 Linux 稳定包准备中', 'First stable Linux package in preparation')}</span>
<pre><code>curl -fsSL https://hiroute.ai/install.sh | sh</code></pre>
diff --git a/apps/website/src/components/HomePage.astro b/apps/website/src/components/HomePage.astro
index bcdce75f..4c9a5402 100644
--- a/apps/website/src/components/HomePage.astro
+++ b/apps/website/src/components/HomePage.astro
@@ -22,64 +22,65 @@ const releaseKinds = [preferredArtifact && t('macOS 桌面应用', 'macOS Deskto
<h1>{en ? <><span>Intelligent routing</span><span>for long-horizon agents.</span></> : <><span>为长程 Agent 而生的</span><span>智能路由引擎。</span></>}</h1>
<p class="intro">{t('一段长任务,不必始终依赖同一个模型。按任务选择 Agent,再根据执行表现与后续工作的需要,决定继续使用还是切换模型。', 'A long task need not rely on one model from start to finish. Match the agent to the task, then choose the right model for each stage—guided by past performance and the work ahead.')}</p>
<a class="jev-callout" href="#how"><strong>Jev</strong><span>{t('官方决策扩展 · 一次调用,选择与评分', 'Official extension · One call to choose and assess')}</span><span aria-hidden="true">↗</span></a>
- <div class="actions"><a class="button" href="#boundaries">{t('看懂长程路由', 'Explore the mechanism')} <span>↓</span></a><a class="button outline" href={en ? '/en/download/' : '/download/'}>{t('下载 HiRoute', 'Download HiRoute')} <span>↗</span></a></div>
+ <div class="actions"><a class="button" href={en ? '/en/download/' : '/download/'}>{t('下载 HiRoute', 'Download HiRoute')} <span>↗</span></a><a class="button outline" href="#boundaries">{t('了解路由机制', 'Explore routing')} <span>↓</span></a></div>
<p class="small-note">{t('开发、研究、分析与自动化。保留惯用工作流,按需投入算力。', 'Code, research, analysis, and automation. Keep your workflow and spend compute where it counts.')}</p>
</div>
- <div class="engine-map">
- <div class="route-top"><span>{t('一个任务,多次决策', 'ONE TASK. MANY DECISIONS.')}</span><span>{t('机制示意', 'MECHANISM')}</span></div>
- <div class="engine-goal"><span>{t('一个长程目标', 'One long-horizon goal')}</span><strong>{t('理解 → 规划 → 执行 → 验证', 'Understand → Plan → Execute → Verify')}</strong></div>
- <div class="engine-link" aria-hidden="true">↓</div>
- <div class="engine-agent"><span>{t('主 Agent / 执行 Agent', 'Main agent / Execution agent')}</span><small>{t('按用途选择工作计划,保留惯用工作流', 'Choose a work plan by purpose; keep the familiar workflow')}</small></div>
- <div class="engine-link" aria-hidden="true">↓</div>
- <div class="engine-center"><img src="/brand/app-icon.svg" alt=""><div><strong>HiRoute</strong><span>{t('模型路由 · 预设思考档位 · 有界接力', 'Model routing · Reasoning profiles · Bounded failover')}</span></div></div>
- <div class="engine-sources"><span>{t('订阅', 'Subscriptions')}</span><span>API</span><span>{t('免费模型', 'Free models')}</span><span>{t('本地模型', 'Local models')}</span></div>
- <div class="engine-feedback">↶ {t('执行事实,成为下一次判断的依据', 'Execution facts inform the next decision')}</div>
+ <div class="journey-art" aria-label={t('一个长任务中的模型路由示意', 'Model routing across one long task')}>
+ <div class="journey-heading"><img src="/brand/mark-switch.svg" alt=""><span>{t('一个任务,多次恰当选择', 'One task. The right model at each stage.')}</span></div>
+ <div class="journey-goal"><span>{t('任务目标', 'THE GOAL')}</span><strong>{t('从想法,到可验证的结果。', 'From an idea to a verified result.')}</strong></div>
+ <div class="journey-stage economy"><span class="journey-dot"></span><div><small>{t('阶段 01 · 明确、局部的工作', 'STAGE 01 · BOUNDED WORK')}</small><h3>{t('经济模型,持续推进', 'Economy model, steady progress')}</h3><p>{t('理解问题 → 调用工具 → 验证结果', 'Understand → Use tools → Verify')}</p><div class="prefix-bars" aria-hidden="true"><i></i><i></i><i></i><i></i><i></i><i></i><i></i></div><span class="journey-cache">{t('阶段内持续复用前缀', 'Reuse the prefix throughout the stage')}</span></div></div>
+ <a class="journey-pivot" href="#how"><span>↻</span><div><strong>{t('上下文压缩 · 重新判断', 'Context compaction · Reassess')}</strong><small>{t('回看执行表现,判断后续工作', 'Assess past performance and the work ahead')}</small></div><b>↗</b></a>
+ <div class="journey-stage primary"><span class="journey-dot"></span><div><small>{t('阶段 02 · 更深入的推理', 'STAGE 02 · DEEPER REASONING')}</small><h3>{t('主力模型,接续完成', 'Primary model, take it further')}</h3><p>{t('分析难点 → 完成实现 → 检查交付', 'Reason → Implement → Check the result')}</p><div class="prefix-bars" aria-hidden="true"><i></i><i></i><i></i><i></i></div><span class="journey-cache">{t('重建新前缀,随后持续复用', 'Build a new prefix, then keep reusing it')}</span></div></div>
+ <p class="journey-footnote">{t('机制示例 · 重新判断也可以继续原模型', 'Illustration · Reassessment can also keep the same model')}</p>
</div>
</section>
<div class="trust-line wrap"><span><b>{t('更省', 'Efficient')}</b> · {t('按需投入算力', 'Compute where it counts')}</span><span><b>{t('更稳', 'Reliable')}</b> · {t('有边界地接力', 'Bounded failover')}</span><span><b>{t('更懂', 'Adaptive')}</b> · {t('用执行事实改进选择', 'Decisions grounded in execution')}</span></div>
<section class="section wrap lifecycle-section" id="boundaries">
- <div class="section-title"><div><div class="eyebrow section-label">{t('01 / 长任务不只做一次模型选择', '01 / A long task is not one model choice')}</div><h2>{t('阶段内保持稳定,', 'Stay stable within a stage.')}<br>{t('在自然边界重新选择。', 'Choose again at natural boundaries.')}</h2></div><p>{t('不必等用户再次提问。Agent 自主执行时,ContextHold 允许切换的上下文重建也能成为决策机会;持续追加的工具续轮保持稳定。', 'A new user prompt is not the only boundary. A ContextHold boundary can also trigger a decision, while ordinary tool continuations stay stable.')}</p></div>
+ <div class="section-title"><div><div class="eyebrow section-label">{t('长任务不只做一次模型选择', 'A long task is not one model choice')}</div><h2>{t('阶段内保持稳定,', 'Stay stable within a stage.')}<br>{t('在自然边界重新选择。', 'Choose again at natural boundaries.')}</h2></div><p>{t('不必等用户再次提问。长会话压缩并重建上下文时,也能成为重新选模的自然机会;普通工具续轮保持当前模型。', 'A new user prompt is not the only boundary. When a long session compacts and rebuilds its context, HiRoute gets another natural chance to choose; ordinary tool continuations stay on the current model.')}</p></div>
<div class="timeline-legend"><span class="legend-economy">{t('经济模型', 'Economy model')}</span><span class="legend-primary">{t('主力模型', 'Primary model')}</span><span>{t('选择决策节点查看接续方式 · 机制示例', 'Select a decision point to inspect the handoff · Illustration')}</span></div>
<div class="timeline-grid">
<div class="turn-caption turn-one">{t('用户提出目标 → Agent 自主推进', 'User sets the goal → Agent works autonomously')}<span>{t('同一个用户任务可以跨多个路由执行轮次', 'One user task can span multiple routing turns')}</span></div>
<div class="turn-caption turn-two">{t('用户补充反馈 → 继续推进', 'User feedback → Continue')}<span>{t('新输入带来新的决策机会', 'New input creates another decision point')}</span></div>
<article class="execution-window first-window"><span class="eyebrow">{t('执行轮次 01', 'ROUTING TURN 01')}</span><h3>{t('经济模型 A', 'Economy model A')}</h3><p>{t('处理明确、局部的工作', 'Handles bounded, local work')}</p><div class="tool-steps"><span>{t('回答', 'Answer')}</span><i>→</i><span>{t('工具', 'Tools')}</span><i>→</i><span>{t('继续', 'Continue')}</span></div><div class="cache-ribbon cache-a"><b>{t('A 的前缀持续追加', 'A’s prefix keeps growing')}</b><span>{t('连续工具往返,保持模型稳定', 'Tool round-trips keep the model stable')}</span></div></article>
- <button class="decision-boundary compact-boundary" type="button" data-boundary="compaction" aria-pressed="true" aria-controls="boundary-detail"><span class="boundary-symbol" aria-hidden="true">↻</span><strong>{t('上下文重建', 'Context rebuilt')}</strong><span class="boundary-status">{t('重新决策', 'DECIDE AGAIN')}</span><small>{t('回看执行表现', 'Assess prior work')}<br>{t('判断后续工作', 'Judge what comes next')}</small></button>
- <article class="execution-window second-window"><span class="eyebrow">{t('执行轮次 02', 'ROUTING TURN 02')}</span><h3>{t('转交主力模型 B', 'Hand off to primary B')}</h3><p>{t('示例:后续工作需要更强推理', 'Example: the next stage needs deeper reasoning')}</p><div class="tool-steps"><span>{t('回答', 'Answer')}</span><i>→</i><span>{t('工具', 'Tools')}</span><i>→</i><span>{t('继续', 'Continue')}</span></div><div class="cache-ribbon cache-b"><b>{t('在 B 上建立新前缀', 'Build a new prefix on B')}</b><span>{t('利用重建时机,再持续复用', 'Switch at rebuild, then reuse consistently')}</span></div></article>
+ <button class="decision-boundary compact-boundary" type="button" data-boundary="compaction" aria-pressed="true" aria-controls="boundary-detail"><span class="boundary-symbol" aria-hidden="true">↻</span><strong>{t('上下文压缩', 'Context compaction')}</strong><span class="boundary-status">{t('重新决策', 'DECIDE AGAIN')}</span><small>{t('回看执行表现', 'Assess prior work')}<br>{t('判断后续工作', 'Judge what comes next')}</small></button>
+ <article class="execution-window second-window"><span class="eyebrow">{t('执行轮次 02', 'ROUTING TURN 02')}</span><h3>{t('转交主力模型 B', 'Hand off to primary B')}</h3><p>{t('示例:后续工作需要更强推理', 'Example: the next stage needs deeper reasoning')}</p><div class="tool-steps"><span>{t('回答', 'Answer')}</span><i>→</i><span>{t('工具', 'Tools')}</span><i>→</i><span>{t('继续', 'Continue')}</span></div><div class="cache-ribbon cache-b"><b>{t('在 B 上建立新前缀', 'Build a new prefix on B')}</b><span>{t('利用压缩后的重建时机,再持续复用', 'Switch after compaction, then reuse consistently')}</span></div></article>
<button class="decision-boundary human-boundary" type="button" data-boundary="human" aria-pressed="false" aria-controls="boundary-detail"><span class="boundary-symbol" aria-hidden="true">+</span><strong>{t('用户补充反馈', 'User feedback')}</strong><span class="boundary-status">{t('重新决策', 'DECIDE AGAIN')}</span><small>{t('新问题或新要求', 'New question or requirement')}<br>{t('也可以继续原模型', 'The same model may continue')}</small></button>
<article class="execution-window third-window"><span class="eyebrow">{t('执行轮次 03', 'ROUTING TURN 03')}</span><h3>{t('继续使用模型 B', 'Continue with model B')}</h3><p>{t('示例:表现胜任,接着完成', 'Example: performance is competent')}</p><div class="tool-steps"><span>{t('回答', 'Answer')}</span><i>→</i><span>{t('工具', 'Tools')}</span><i>→</i><span>{t('继续', 'Continue')}</span></div><div class="cache-ribbon cache-b"><b>{t('B 的前缀继续追加', 'B’s prefix keeps growing')}</b><span>{t('重新判断不代表必须切换', 'Reassessment does not force a switch')}</span></div></article>
</div>
- <div class="boundary-detail" id="boundary-detail" aria-live="polite"><div><span class="detail-kicker" id="boundary-kicker">{t('自主执行中的决策机会', 'A decision point during autonomous work')}</span><h3 id="boundary-title">{t('重建上下文时,可以重新选择。', 'A rebuilt context is a chance to choose again.')}</h3><p id="boundary-copy">{t('ContextHold 判断当前边界允许切换时,HiRoute 重新评估下一段工作;不要求客户端额外报告“发生了压缩”。', 'When ContextHold permits a switch, HiRoute reassesses the work ahead; the client need not send a separate compaction signal.')}</p></div><div class="boundary-outcome"><span>{t('这一节点的示例', 'At this point')}</span><p><b>{t('评估上一阶段', 'Assess the prior stage')}</b><span id="boundary-assessment">{t('回看模型 A 已完成的执行', 'Assess the work already performed by model A')}</span></p><p><b>{t('选择下一分支', 'Choose the next branch')}</b><span id="boundary-choice">{t('选择主力分支,让模型 B 接续', 'Choose the primary branch; model B takes over')}</span></p></div></div>
+ <div class="boundary-detail" id="boundary-detail" aria-live="polite"><div><span class="detail-kicker" id="boundary-kicker">{t('自主执行中的决策机会', 'A decision point during autonomous work')}</span><h3 id="boundary-title">{t('上下文压缩后,可以重新选择。', 'After context compaction, choose again.')}</h3><p id="boundary-copy">{t('长会话压缩并重建上下文时,可复用前缀本来就会变化。HiRoute 利用这个时机评估上一阶段和后续工作;客户端无需额外识别或上报压缩事件。', 'Compaction rebuilds a long session around a new prefix. HiRoute uses that moment to assess the prior stage and the work ahead; the client does not need to detect or report compaction separately.')}</p></div><div class="boundary-outcome"><span>{t('这一节点的示例', 'At this point')}</span><p><b>{t('评估上一阶段', 'Assess the prior stage')}</b><span id="boundary-assessment">{t('回看模型 A 已完成的执行', 'Assess the work already performed by model A')}</span></p><p><b>{t('选择下一分支', 'Choose the next branch')}</b><span id="boundary-choice">{t('选择主力分支,让模型 B 接续', 'Choose the primary branch; model B takes over')}</span></p></div></div>
</section>
- <section class="section wrap" id="product"><div class="section-title"><div><div class="eyebrow section-label">{t('02 / 先选择任务路径,再配置算力', '02 / Route the task. Then the compute.')}</div><h2>{t('按任务选择 Agent,', 'Choose the agent for the task.')}<br>{t('按需要配置模型与算力。', 'Then configure models and compute.')}</h2></div><p>{t('面向开发、研究、复杂分析与通用自动化,不把产品定义绑定在某个 Agent 上。', 'Designed for coding, research, deep analysis, and general automation—not around one hard-coded agent.')}</p></div><div class="features">
- <article class="feature"><span class="num">01—</span><h3>{t('按用途选择执行 Agent', 'Choose an agent by purpose')}</h3><p>{t('主 Agent 发现开放的工作计划,按目标委派任务并取得结果。任务委派与模型路由可独立启用。', 'A main agent discovers allowed work plans, delegates a goal, and retrieves the result. Delegation and model routing remain independent.')}</p></article>
- <article class="feature"><span class="num">02—</span><h3>{t('按需要配置模型与算力', 'Configure models and compute')}</h3><p>{t('组合模型来源、思考档位和候选顺序;可以智能省钱、固定模型或免费优先。', 'Compose model sources, reasoning profiles, and ordered candidates; use smart saving, fixed, or free-first routing.')}</p></article>
- <article class="feature"><span class="num">03—</span><h3>{t('让接力有边界', 'Keep failover bounded')}</h3><p>{t('候选满足工具、图像和上下文要求;候选耗尽时明确停止,开始交付后不透明换模。', 'Candidates satisfy tool, image, and context requirements; exhaustion stops explicitly and delivery is not silently switched.')}</p></article>
- </div></section>
- <section class="night" id="how"><div class="section wrap mechanism"><div><div class="eyebrow section-label">{t('03 / 由 TypeSafe Jev 提供决策能力', '03 / Powered by TypeSafe Jev')}</div><span class="jev-extension-label">{t('HiRoute 官方决策扩展', 'Official HiRoute decision extension')}</span><h2>{t('用 Jev 选择模型,', 'Let Jev choose the model.')}<br><em>{t('也评价之前的表现。', 'And assess the work so far.')}</em></h2><p class="explain">{t('同一次上游调用选择下一分支,并可对上一阶段给出胜任度评分。不是每个工具步骤都调用一次。', 'One upstream call chooses the next branch and can assess the previous stage. It is not called for every tool step.')}</p><p class="policy-principle">{t('胜任度只负责阻止冒险降本,', 'Competence blocks risky cost-cutting.')}<br>{t('复杂度负责提供降本机会。', 'Complexity creates opportunities to save.')}</p><p class="explain">{t('规则模式(rules)按简单概率与胜任度门槛执行;自动模式(auto)综合任务需要和历史表现直接选择。两种模式二选一。', 'Rules applies a simple-task probability and competence floor; auto chooses from task needs and prior performance. Choose one mode.')}</p><a class="button accent" href={en ? '/en/docs/jev-decider/' : '/docs/jev-decider/'}>{t('部署官方 Jev 扩展', 'Deploy the official Jev extension')} ↗</a><p class="jev-optional">{t('Jev 是可选扩展;也可使用内置规则或自建决策服务。', 'Jev is optional; built-in rules and custom decision services remain available.')}</p></div><div class="stage-track">
- <article class="stage"><span>01</span><div><span class="primitive-label">Score / {t('回看', 'LOOK BACK')}</span><h3>{t('之前的模型,是否胜任?', 'Was the previous model up to the job?')}</h3><p>{t('结合接受的回答、工具状态和用户反馈评价已经发生的执行。', 'Assess completed work from accepted output, tool outcomes, and user feedback.')}</p></div></article>
- <article class="stage"><span>02</span><div><span class="primitive-label">Choice / {t('选择', 'LOOK AHEAD')}</span><h3>{t('接下来的工作,该用哪个分支?', 'Which branch should handle the next stage?')}</h3><p>{t('auto 直接选允许分支;rules 使用简单/复杂概率并以胜任度限制冒险降本。', 'Auto chooses directly; rules uses simple/complex probability with competence as a guardrail.')}</p></div></article>
- <article class="stage"><span>03</span><div><span class="primitive-label">HiRoute / {t('执行与记录', 'EXECUTE AND RETAIN')}</span><h3>{t('让选择生效,让评价有归属', 'Apply the choice and attribute the score')}</h3><p>{t('HiRoute 执行分支,把评分记到此前实际执行的模型阶段。', 'HiRoute executes the branch and records the score against the model’s prior stage.')}</p><a class="text-link" href={en ? '/en/docs/decision-api/' : '/docs/decision-api/'}>{t('查看决策协议', 'Explore the decision contract')} →</a></div></article>
- </div></div></section>
+
+ <section class="night" id="how"><div class="section wrap jev-section">
+ <div class="jev-heading"><div><span class="jev-extension-label">{t('HiRoute 官方决策扩展', 'Official HiRoute decision extension')}</span><h2>{t('Jev,一次判断。', 'Jev. One decision.')}<br><em>{t('回看表现,也看向下一步。', 'Look back. Choose what comes next.')}</em></h2></div><div><p class="explain">{t('结合当前任务与执行历史,在一次上游请求中选择下一分支,并可评价此前模型的胜任度。', 'Combine the current task with execution history. One upstream request chooses the next branch and can assess the previous model’s competence.')}</p><a class="text-link" href={en ? '/en/docs/jev-decider/' : '/docs/jev-decider/'}>{t('部署官方 Jev 扩展', 'Deploy the Jev extension')} ↗</a></div></div>
+ <p class="diagram-mobile-hint">{t('左右滑动查看完整机制图', 'Swipe to explore the full diagram')}</p><figure class="jev-diagram" tabindex="0" aria-label={t('Jev 决策机制图,可横向滚动', 'Jev decision diagram, horizontally scrollable')}><img src={`/decision-assets/jev-decision-${en ? 'en' : 'zh-CN'}.svg`} alt={t('HiRoute 提供当前输入、执行历史与允许分支;Jev 单次调用返回下一分支与可选胜任度;HiRoute 执行并记录。', 'HiRoute supplies current input, history and allowed branches. A single Jev call returns a branch and optional competence assessment. HiRoute executes and records.')} loading="lazy"></figure>
+ <div class="jev-modes"><article><span>{t('复杂度 · 寻找降本机会', 'COMPLEXITY · FIND ROOM TO SAVE')}</span><h3>{t('接下来的任务,经济模型能做吗?', 'Can an economy model handle what comes next?')}</h3><p>{t('Jev 给出任务简单的概率。达到设定门槛,才考虑使用经济分支;复杂任务继续交给主力分支。', 'Jev estimates the probability that the task is simple. Only consider economy when it meets your threshold; complex work stays on primary.')}</p></article><article><span>{t('胜任度 · 守住执行质量', 'COMPETENCE · PROTECT QUALITY')}</span><h3>{t('此前的表现,是否值得继续降本?', 'Does prior performance support saving?')}</h3><p>{t('本次对上一阶段的评分低于门槛时,阻止降本。高分不能抵消任务复杂度;无合法评分时,仅依据复杂度判断。', 'A low assessment of the previous stage blocks cost-cutting. A high score cannot override task complexity; without a valid score, complexity alone decides.')}</p></article></div>
+ <div class="jev-bottom"><p>{t('胜任度只负责阻止冒险降本,复杂度负责提供降本机会。', 'Competence blocks risky cost-cutting; complexity creates opportunities to save.')}</p><small>{t('以上为官方扩展的 Rules 策略 · 阈值可配置 · Jev 为可选扩展', 'Rules policy in the official extension · Configurable thresholds · Jev is optional')}</small></div>
+ </div></section>
<section class="section wrap" id="performance">
- <div class="section-title"><div><div class="eyebrow section-label">{t('04 / 为下一次决策保留证据', '04 / Evidence for the next decision')}</div><h2>{t('不只知道选了谁,', 'See more than who was chosen.')}<br>{t('也看得见做得怎样。', 'See how the work went.')}</h2></div><p>{t('查看当前计划所选模型一段时间内的阶段胜任度,无需输入内部模型 ID。', 'Review stage competence over time for models selected by the current plan—without typing internal model IDs.')}</p></div>
+ <div class="section-title"><div><div class="eyebrow section-label">{t('为下一次决策保留证据', 'Evidence for the next decision')}</div><h2>{t('不只知道选了谁,', 'See more than who was chosen.')}<br>{t('也看得见做得怎样。', 'See how the work went.')}</h2></div><p>{t('查看当前计划所选模型一段时间内的阶段胜任度,无需输入内部模型 ID。', 'Review stage competence over time for models selected by the current plan—without typing internal model IDs.')}</p></div>
<figure class="product-shot"><img src={`/decision-assets/quality-${en ? 'en' : 'zh-CN'}.png`} alt={t('HiRoute 运行表现界面,包含阶段评分、未评分和部分历史', 'HiRoute runtime performance with stage scores, unrated work, and partial history')} loading="lazy"></figure>
<p class="caption">{t('真实产品组件 · 演示数据,非模型评测。保留未评分、部分历史和评分尚未覆盖最新轮次等实际状态。', 'Real product components · Illustrative data, not a model benchmark. Unrated, partial, and not-yet-covered turns remain explicit.')}</p>
<div class="proof-bottom"><article><h3>{t('分数始终带着范围', 'A score keeps its scope')}</h3><p>{t('评分属于具体计划、版本、模型和执行阶段;旧评分不会假装覆盖新增轮次。', 'A score belongs to a plan, revision, model, and execution stage; new turns are not silently covered by an older assessment.')}</p></article><article><h3>{t('证据可以帮助细分计划', 'Evidence can refine work plans')}</h3><p>{t('主 Agent 可查询高低分任务,结合证据判断是否需要建立更聚焦的工作场景。', 'A main agent can query high- and low-scoring tasks, inspect evidence, and propose a narrower work plan.')}</p></article></div>
- <div class="specialization"><div class="specialization-heading"><span class="eyebrow">{t('从执行证据到更合适的委派', 'From evidence to better delegation')}</span><h3>{t('任务更聚焦,模型就有更合适的位置。', 'Focused work gives each model a clearer role.')}</h3><p>{t('先看多次执行证据,再决定是否拆出更专门的场景。', 'Inspect repeated execution evidence before proposing a specialized scenario.')}</p></div><div class="specialization-flow"><article><span class="step-number">01 / {t('观察', 'OBSERVE')}</span><h4>{t('同一模型,不同任务表现不同', 'Performance varies by task')}</h4><p>{t('先按计划、版本和任务范围查看阶段证据。', 'Inspect stages by plan, revision, and task scope.')}</p></article><span class="flow-arrow">→</span><article><span class="step-number">02 / {t('细分', 'SPECIALIZE')}</span><h4>{t('明确更聚焦的工作场景', 'Define a narrower scenario')}</h4><p>{t('缩小范围,明确输入与验证条件。', 'Narrow the scope and make acceptance criteria explicit.')}</p></article><span class="flow-arrow">→</span><article><span class="step-number">03 / {t('委派', 'DELEGATE')}</span><h4>{t('把任务交给适合的计划', 'Delegate to the fitting plan')}</h4><p>{t('再观察后续表现,不凭单次低分自动改计划。', 'Observe later performance; one low score never rewrites a plan automatically.')}</p></article></div></div>
+
</section>
+ <section class="section wrap" id="product"><div class="section-title"><div><div class="eyebrow section-label">{t('先选择任务路径,再配置算力', 'Route the task. Then the compute.')}</div><h2>{t('按任务选择 Agent,', 'Choose the agent for the task.')}<br>{t('按需要配置模型与算力。', 'Then configure models and compute.')}</h2></div><p>{t('面向开发、研究、复杂分析与通用自动化,不把产品定义绑定在某个 Agent 上。', 'Designed for coding, research, deep analysis, and general automation—not around one hard-coded agent.')}</p></div><div class="features">
+ <article class="feature"><span class="num">01—</span><h3>{t('按用途选择执行 Agent', 'Choose an agent by purpose')}</h3><p>{t('主 Agent 发现开放的工作计划,按目标委派任务并取得结果。任务委派与模型路由可独立启用。', 'A main agent discovers allowed work plans, delegates a goal, and retrieves the result. Delegation and model routing remain independent.')}</p></article>
+ <article class="feature"><span class="num">02—</span><h3>{t('按需要配置模型与算力', 'Configure models and compute')}</h3><p>{t('组合模型来源、思考档位和候选顺序;可以智能省钱、固定模型或免费优先。', 'Compose model sources, reasoning profiles, and ordered candidates; use smart saving, fixed, or free-first routing.')}</p></article>
+ <article class="feature"><span class="num">03—</span><h3>{t('让接力有边界', 'Keep failover bounded')}</h3><p>{t('候选满足工具、图像和上下文要求;候选耗尽时明确停止,开始交付后不透明换模。', 'Candidates satisfy tool, image, and context requirements; exhaustion stops explicitly and delivery is not silently switched.')}</p></article>
+ </div></section>
+
<section class="section wrap ecosystem-section" id="ecosystem">
- <div class="section-title"><div><div class="eyebrow section-label">{t('05 / Agent 生态', '05 / Agent ecosystem')}</div><h2>{t('不绑定一种 Agent,', 'Not tied to one agent.')}<br>{t('也不局限于编程。', 'Not limited to coding.')}</h2></div><p>{t('集成状态按真实支持范围开放;当前可用与规划方向明确分开。', 'Integration status follows real support; available and planned paths stay visibly separate.')}</p></div>
+ <div class="section-title"><div><div class="eyebrow section-label">{t('Agent 生态', 'Agent ecosystem')}</div><h2>{t('不绑定一种 Agent,', 'Not tied to one agent.')}<br>{t('也不局限于编程。', 'Not limited to coding.')}</h2></div><p>{t('集成状态按真实支持范围开放;当前可用与规划方向明确分开。', 'Integration status follows real support; available and planned paths stay visibly separate.')}</p></div>
<div class="ecosystem-groups"><article><h3>{t('当前适配', 'Available now')}</h3><div class="agent-names"><span>Codex</span><span>Claude Code</span></div></article><article class="ecosystem-planned"><h3>{t('后续方向', 'Planned')}</h3><div class="agent-names"><span>Qoder</span><span>Pi</span><span>DeepSeek Harness</span><span>OpenCode</span><span>Hermes Agent</span><span>OpenClaw</span></div><p>{t('具体顺序按真实适配和验证推进。', 'Order depends on real integration and validation.')}</p></article></div>
</section>
<section class="section wrap extension-section" id="extensions">
- <div><div class="eyebrow section-label">{t('06 / 一套接口,接入你的策略', '06 / Your policy, one interface')}</div><h2>{t('把你的判断方式,', 'Your decision policy.')}<br>{t('接入 HiRoute。', 'One HiRoute interface.')}</h2><p>{t('用 Jev、LLM 或自己的策略实现决策服务。通用协议支持允许分支选择与可选胜任度评分。', 'Build a decision service with Jev, an LLM, or your own policy. The general API supports allowed-branch choice and optional competence scoring.')}</p><a class="text-link" href={en ? '/en/docs/decision-api/' : '/docs/decision-api/'}>{t('决策 API 文档', 'Decision API documentation')} →</a><br><a class="text-link" href="/api/decision.openapi.json" download>{t('下载 OpenAPI', 'Download OpenAPI')} ↓</a></div>
+ <div><div class="eyebrow section-label">{t('一套接口,接入你的策略', 'Your policy, one interface')}</div><h2>{t('把你的判断方式,', 'Your decision policy.')}<br>{t('接入 HiRoute。', 'One HiRoute interface.')}</h2><p>{t('用 Jev、LLM 或自己的策略实现决策服务。通用协议支持允许分支选择与可选胜任度评分。', 'Build a decision service with Jev, an LLM, or your own policy. The general API supports allowed-branch choice and optional competence scoring.')}</p><a class="text-link" href={en ? '/en/docs/decision-api/' : '/docs/decision-api/'}>{t('决策 API 文档', 'Decision API documentation')} →</a><br><a class="text-link" href="/api/decision.openapi.json" download>{t('下载 OpenAPI', 'Download OpenAPI')} ↓</a></div>
<div class="code-box"><div class="code-heading">{t('决策 API / 单次请求', 'DECISION API / SINGLE REQUEST')}</div><pre><code><span class="accent">POST</span> /v1/decisions
&#123;
@@ -89,17 +90,16 @@ const releaseKinds = [preferredArtifact && t('macOS 桌面应用', 'macOS Deskto
</section>
<section class="section wrap" id="start">
- <div class="download-row"><div><div class="eyebrow section-label">{t('07 / 从 HiRoute 开始', '07 / Start with HiRoute')}</div><h2>{t('保留工作流,', 'Keep your workflow.')}<br>{t('重新安排算力。', 'Rearrange the compute.')}</h2><p>{release ? t(`稳定版本 ${release.version} 已通过统一发布清单提供 ${releaseKinds}。`, `Stable ${release.version} serves ${releaseKinds} through one release manifest.`) : t('公开版本正在准备中。macOS 使用桌面应用,Linux 可通过一条命令安装完整的无界面管理端。', 'The first public release is being prepared. Use Desktop on macOS or install the complete headless management surface on Linux with one command.')}</p><div class="actions"><a class="button" href={en ? '/en/download/' : '/download/'}>{t('选择安装方式', 'Choose an install path')} ↗</a><a class="button outline" href={en ? '/en/docs/' : '/docs/'}>{t('阅读快速开始', 'Read the Quickstart')} ↗</a></div></div>
+ <div class="download-row"><div><div class="eyebrow section-label">{t('从 HiRoute 开始', 'Start with HiRoute')}</div><h2>{t('保留工作流,', 'Keep your workflow.')}<br>{t('重新安排算力。', 'Rearrange the compute.')}</h2><p>{release ? t(`稳定版本 ${release.version} 现已提供 ${releaseKinds}。`, `Stable ${release.version} is available for ${releaseKinds}.`) : t('公开版本正在准备中。macOS 使用桌面应用,Linux 可通过一条命令安装完整的无界面管理端。', 'The first public release is being prepared. Use Desktop on macOS or install the complete headless management surface on Linux with one command.')}</p><div class="actions"><a class="button" href={en ? '/en/download/' : '/download/'}>{t('选择安装方式', 'Choose an install path')} ↗</a><a class="button outline" href={en ? '/en/docs/' : '/docs/'}>{t('阅读快速开始', 'Read the Quickstart')} ↗</a></div></div>
<div class="download-state">{release && linuxArtifact ? <><b>{t('Linux 无界面版', 'Linux headless')} · {release.version}</b><p>{linuxArtifact.architecture} · {formatBytes(linuxArtifact.size, language)}</p><code>curl -fsSL https://hiroute.ai/install.sh | sh</code><a class="text-link" href={en ? '/en/docs/install-linux/' : '/docs/install-linux/'}>{t('查看安装与启动步骤', 'Installation and startup')} →</a><span class="tag">SHA256 {linuxArtifact.sha256.slice(0, 12)}…</span></> : release && preferredArtifact ? <><b>HiRoute {release.version}</b><p>macOS {preferredArtifact.architecture} · {formatBytes(preferredArtifact.size, language)}</p><a class="button download-button" href={artifactUrl(release, preferredArtifact)}>{t('下载桌面应用', 'Download Desktop')}</a><span class="tag">{preferredArtifact.distribution === 'self-signed' ? t('自签名', 'SELF-SIGNED') : 'DEVELOPER ID'} · SHA256 {preferredArtifact.sha256.slice(0, 12)}…</span></> : <><b>{t('桌面应用 + 无界面版', 'Desktop + headless')}</b><p>{t('发布前先完成完整性检查、安装走查和发布审核。', 'Integrity checks, installation review, and release approval come first.')}</p><span class="tag">{t('准备中', 'IN PREPARATION')}</span></>}</div>
</div>
</section>
- <section class="section wrap faq-section" id="faq"><div class="eyebrow section-label">{t('08 / 常见问题', '08 / Questions')}</div><h2>{t('常见问题', 'Frequently asked questions')}</h2>
- <details><summary>{t('模型切换会破坏 KV cache 吗?', 'Does model switching destroy the KV cache?')}</summary><p>{t('不同模型不能共享 KV cache。HiRoute 在 ContextHold 允许重选的自然边界做决定;阶段内连续工具续轮保持模型稳定。实际命中仍取决于模型服务、配置和缓存生命周期。', 'Different models do not share KV cache. HiRoute decides at natural boundaries where ContextHold permits reselection; ordinary tool continuations stay on the same model. Actual hits still depend on provider, configuration, and cache lifetime.')}</p></details>
- <details id="history-faq"><summary>{t('上下文重建后,决策服务还能看到历史吗?', 'Can the decision service still see history after context rebuild?')}</summary><p>{t('可以看到 HiRoute 独立维护的简化 Agent 轮次历史:用户问题、接受的回答、工具名称与粗粒度状态、执行分支。它不是完整对话转录,淘汰或重启后可标记为部分历史。', 'It receives HiRoute’s independent simplified Agent-turn history: user input, accepted output, tool names and coarse outcomes, and executed branches. It is not a full transcript and can be marked partial after eviction or restart.')}</p></details>
+ <section class="section wrap faq-section" id="faq"><div class="eyebrow section-label">{t('常见问题', 'Questions')}</div><h2>{t('常见问题', 'Frequently asked questions')}</h2>
+ <details><summary>{t('模型切换会破坏 KV cache 吗?', 'Does model switching destroy the KV cache?')}</summary><p>{t('不同模型不能共享 KV cache。HiRoute 尽量把换模放在上下文压缩后的自然重建时机;普通工具续轮保持模型稳定。实际命中仍取决于模型服务、配置和缓存生命周期。', 'Different models do not share KV cache. HiRoute aligns model changes with natural rebuilds after context compaction when possible, while ordinary tool continuations stay on the same model. Actual hits still depend on provider, configuration, and cache lifetime.')}</p></details>
+ <details id="history-faq"><summary>{t('上下文压缩后,决策服务还能看到历史吗?', 'Can the decision service still see history after compaction?')}</summary><p>{t('可以看到 HiRoute 独立维护的简化 Agent 轮次历史:用户问题、接受的回答、工具名称与粗粒度状态、执行分支。它不是完整对话转录,淘汰或重启后可标记为部分历史。', 'It receives HiRoute’s independent simplified Agent-turn history: user input, accepted output, tool names and coarse outcomes, and executed branches. It is not a full transcript and can be marked partial after eviction or restart.')}</p></details>
<details><summary>{t('一定要部署 Jev 吗?', 'Must I deploy Jev?')}</summary><p>{t('不需要。内置规则可以开始路由;Jev 是官方可选扩展,也可以按同一 Decision API 实现自己的服务。', 'No. Built-in rules are enough to start; Jev is an optional official extension, and you can implement your own service against the same Decision API.')}</p></details>
<details><summary>{t('胜任度评分是不是成功率?', 'Is competence a success rate?')}</summary><p>{t('不是。它表达模型对具体执行阶段的胜任程度,必须结合计划、版本、评分覆盖范围和部分证据状态理解。', 'No. It represents competence for a specific execution stage and must be interpreted with plan, revision, assessment coverage, and partial-evidence state.')}</p></details>
- <details><summary>{t('安装包从哪里下载?', 'Where are installers served from?')}</summary><p>{t('下载链接直接指向 hiroute.ai/releases/<版本>/<文件名> 的不可变 OSS 对象;不存在的文件不会回退到首页。', 'Links point directly to immutable OSS objects under hiroute.ai/releases/<version>/<filename>; missing files never fall back to the homepage.')}</p></details>
</section>
<script is:inline src="/scripts/home.js" defer></script>
</SiteLayout>
diff --git a/apps/website/src/layouts/SiteLayout.astro b/apps/website/src/layouts/SiteLayout.astro
index 875e3f10..83543419 100644
--- a/apps/website/src/layouts/SiteLayout.astro
+++ b/apps/website/src/layouts/SiteLayout.astro
@@ -5,6 +5,7 @@ const {
fullWidth = false,
} = Astro.props;
const english = language === 'en';
+const githubUrl = 'https://github.com/higress-group/HiRoute';
const canonical = new URL(path, Astro.site);
const alternate = new URL(alternatePath, Astro.site);
const zhPath = english ? alternatePath : path;
@@ -48,6 +49,7 @@ const nav = english ? [
</a>
<nav class="nav" aria-label={english ? 'Primary navigation' : '主要导航'}>
{nav.map(([label, href]) => <a href={href}>{label}</a>)}
+ <a class="nav-github" href={githubUrl} target="_blank" rel="noreferrer">GitHub <span aria-hidden="true">↗</span></a>
<div class="language-switch" aria-label={english ? 'Language' : '语言'}>
<a data-language lang="zh-CN" hreflang="zh-CN" href={zhPath} aria-current={!english ? 'page' : undefined}>中文</a>
<a data-language lang="en" hreflang="en" href={enPath} aria-current={english ? 'page' : undefined}>EN</a>
@@ -64,6 +66,7 @@ const nav = english ? [
<div class="footer-links">
<a href={english ? '/en/docs/' : '/docs/'}>{english ? 'Docs' : '文档'}</a>
<a href={english ? '/en/changelog/' : '/changelog/'}>{english ? 'Changelog' : '更新日志'}</a>
+ <a href={githubUrl} target="_blank" rel="noreferrer">GitHub ↗</a>
</div>
</footer>
</body>
diff --git a/apps/website/src/styles/site.css b/apps/website/src/styles/site.css
index 984cc763..8711f59a 100644
--- a/apps/website/src/styles/site.css
+++ b/apps/website/src/styles/site.css
@@ -17,8 +17,10 @@ section{scroll-margin-top:28px}
.wordmark{display:flex;gap:10px;align-items:center;flex-shrink:0}
.wordmark .brand-symbol{width:39px;height:39px}
.wordmark .brand-lettering{width:142px;height:auto}
-.nav{display:flex;align-items:center;gap:29px;font-size:14px}
+.nav{display:flex;align-items:center;gap:23px;font-size:14px;flex-wrap:wrap}
.nav a:hover{color:#475dd8}
+.nav .nav-github{display:inline-flex;align-items:center;gap:5px;padding:7px 10px;border:1px solid var(--line);border-radius:6px;font-weight:600}
+.nav .nav-github:hover{border-color:#aeb8e8;background:#eef0fb}
.language-switch{display:flex;gap:2px;padding:3px;border:1px solid var(--line);border-radius:6px;flex:none}
.language-switch a{font-size:11px;padding:5px 7px;border-radius:3px;color:var(--muted)}
.language-switch a[aria-current=page]{background:#e6eafb;color:#344ab8;font-weight:600}
@@ -71,8 +73,8 @@ html:lang(en) .section-title>p{flex:0 1 335px}
.scenario-tabs{display:flex;gap:6px;flex-wrap:wrap;padding-top:15px}
.scenario-tabs button{font-size:11px;padding:7px 10px;border-radius:4px;background:transparent;border:1px solid #d2d9e9;color:#64708a}
.scenario-tabs button[aria-pressed=true]{background:var(--white);color:var(--ink);border-color:#a5b0d8}
-.trust-line{display:flex;justify-content:space-between;gap:18px;flex-wrap:wrap;padding:22px 0 25px;border-top:1px solid var(--line);border-bottom:1px solid var(--line);font-size:12px;color:var(--muted)}
-.trust-line b{font-weight:500;color:var(--ink)}
+.trust-line{display:flex;justify-content:space-between;gap:18px;flex-wrap:wrap;padding:26px 0 29px;border-top:1px solid var(--line);border-bottom:1px solid var(--line);font-size:16px;line-height:1.8;color:#596174}
+.trust-line b{font-size:18px;font-weight:600;color:var(--ink)}
.section{padding:88px 0}
.section-title{display:flex;justify-content:space-between;align-items:flex-end;gap:30px}
.section h2{font-size:34px;letter-spacing:-.9px;line-height:1.45}
@@ -427,3 +429,54 @@ html:lang(en) .section-title>p{flex:0 1 335px}
.docs-content blockquote{margin:20px 0;padding:4px 20px;border-left:3px solid var(--highlight)}
@media(max-width:900px){.site-header{height:auto;min-height:90px;flex-wrap:wrap;padding:20px 0}.site-header .nav{width:100%;overflow-x:auto;padding-bottom:5px}.docs-shell{grid-template-columns:minmax(0,1fr);gap:28px}.docs-nav{position:static;border-right:0;border-bottom:1px solid var(--line);padding:0 0 20px;grid-template-columns:repeat(2,minmax(0,1fr))}.docs-nav>span{grid-column:1/-1}.changelog-list article{grid-template-columns:1fr 1fr}.artifact-row{align-items:flex-start}.platform-heading{align-items:start}}
@media(max-width:600px){.page-hero{padding:52px 0 38px}.content-section{padding:38px 0}.release-card,.empty-release{padding:23px}.release-heading,.artifact-row,.platform-heading{display:block}.platform-heading>.text-link{margin-top:17px}.install-command{padding:20px}.artifact-row .button{margin-top:18px;width:100%}.changelog-list article{grid-template-columns:1fr}.docs-shell{padding-top:28px}.docs-content>h1{font-size:31px}.docs-content h2{font-size:22px}.docs-nav{display:flex;gap:6px;overflow-x:auto;margin:0 -18px;padding:0 18px 16px;scrollbar-width:thin}.docs-nav>span{display:none}.docs-nav a{flex:none;white-space:nowrap;border:1px solid var(--line)}.site-header .nav{gap:20px}}
+/* Long-horizon routing overview. */
+.hero{grid-template-columns:1fr 1.05fr;gap:64px;padding:76px 0 68px}
+.hero h1{font-size:clamp(36px,4vw,55px);letter-spacing:-1.4px}
+.hero .intro{max-width:470px;line-height:1.85}
+.journey-art{position:relative;min-width:0;padding:28px 30px 20px;border-radius:24px;background:radial-gradient(ellipse at top right,#e5e5ff,transparent 65%),#f0f2f9;border:1px solid #dde1ee;box-shadow:0 20px 65px #2735780c}
+.journey-heading{display:flex;align-items:center;gap:10px;font-size:12px;color:#56618a}
+.journey-heading img{width:23px;height:23px}
+.journey-goal{padding:25px 0 24px}
+.journey-goal>span{display:block;font-size:10px;color:#6d7690;letter-spacing:1px;margin-bottom:8px}
+.journey-goal strong{font-size:22px;line-height:1.5}
+.journey-stage{position:relative;display:flex;gap:15px;padding:19px 17px;background:#ffffffc9;border-radius:13px;border:1px solid #dce2f4}
+.journey-stage>div{min-width:0}
+.journey-dot{width:10px;height:10px;border-radius:50%;background:#6276dc;flex:none;margin-top:5px;box-shadow:0 0 0 4px #e6eafa}
+.journey-stage small{font-size:10px;letter-spacing:.4px;color:#6477b0}
+.journey-stage h3{font-size:19px;margin:8px 0}
+.journey-stage p{font-size:12px;color:#606b83}
+.prefix-bars{display:flex;gap:4px;height:7px;margin:17px 0 8px}
+.prefix-bars i{width:25px;background:#7184de;border-radius:2px}
+.prefix-bars i:last-child{opacity:.35}
+.journey-cache{font-size:10px;color:#687795}
+.journey-pivot{display:flex;align-items:center;gap:12px;padding:19px 12px;margin-left:20px;border-left:1px dashed #adb7d8}
+.journey-pivot>span{font-size:25px;color:#6b5cbb}
+.journey-pivot strong{display:block;font-size:13px}
+.journey-pivot small{display:block;font-size:11px;color:#69718b;margin-top:5px}
+.journey-pivot>b{margin-left:auto;color:#6b5cbb}
+.journey-pivot:hover strong{color:#584aba}
+.journey-stage.primary{border-color:#e3daef}
+.primary .journey-dot,.primary .prefix-bars i{background:#9674c5}
+.primary small{color:#82639f}
+.journey-footnote{font-size:10px;color:#707991;margin-top:17px;text-align:center}
+.jev-heading{display:grid;grid-template-columns:1.25fr 1fr;gap:70px;align-items:center}
+.jev-heading h2{font-size:clamp(30px,3.3vw,42px);line-height:1.4}
+.jev-heading .explain{font-size:15px;color:#b8c2d8;line-height:1.9;margin-bottom:22px}
+.jev-section .text-link{color:#c5ceff}
+.jev-diagram{margin:36px 0 28px}
+.diagram-mobile-hint{display:none}
+.jev-diagram:focus-visible{outline:2px solid #9aa7ff;outline-offset:5px}
+@media(max-width:760px){.diagram-mobile-hint{display:block;margin-top:28px;color:#aebeff;font-size:12px}}
+.jev-diagram img{width:100%;height:auto}
+.jev-modes{display:grid;grid-template-columns:1fr 1fr;gap:36px}
+.jev-modes article{border-top:1px solid #34405a;padding:24px 0}
+.jev-modes article>span{font-size:12px;color:#aebeff}
+.jev-modes h3{font-size:21px;margin:12px 0}
+.jev-modes p{font-size:14px;color:#b8c2d8;line-height:1.85}
+.jev-bottom{padding-top:20px;border-top:1px solid #34405a}
+.jev-bottom p{font-size:19px;color:#c4ccff}
+.jev-bottom small{display:block;color:#9eacc6;font-size:12px;margin-top:12px;line-height:1.8}
+.product-shot{box-shadow:0 22px 65px #27357812;border-radius:14px;overflow:hidden}
+#product{padding-top:30px}
+@media(max-width:1000px){.hero{gap:30px}.jev-heading{gap:32px}.journey-art{padding:24px 20px}.journey-goal strong{font-size:19px}}
+@media(max-width:760px){.hero{grid-template-columns:1fr;padding:40px 0;gap:35px}.hero h1{font-size:38px;letter-spacing:-1px}.journey-art{padding:24px 20px}.journey-stage h3{font-size:18px}.jev-heading,.jev-modes{grid-template-columns:1fr;gap:20px}.jev-diagram{overflow-x:auto;border-radius:16px}.jev-diagram img{min-width:740px}.jev-bottom p{font-size:17px}.section{padding:58px 0}.jev-modes article{padding:18px 0}.jev-heading h2{font-size:30px}}
diff --git a/decision-extensions/assets/jev-decision-en.svg b/decision-extensions/assets/jev-decision-en.svg
new file mode 100644
index 00000000..00673c6b
--- /dev/null
+++ b/decision-extensions/assets/jev-decision-en.svg
@@ -0,0 +1,7 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="1120" height="660" viewBox="0 0 1120 660" role="img" aria-labelledby="title desc">
+<title id="title">One context. Two perspectives.</title><desc id="desc">One upstream Jev request · Choice and assessment returned together. Assess the previous stage. Choose for the next one.</desc>
+<defs><marker id="arrow" markerWidth="8" markerHeight="8" refX="6" refY="4" orient="auto"><path d="M0 0L7 4L0 8" fill="none" stroke="#9ba9ee" stroke-width="1.5"/></marker><linearGradient id="hub" x2="1" y2="1"><stop stop-color="#303764"/><stop offset="1" stop-color="#222943"/></linearGradient></defs>
+<rect width="1120" height="660" rx="24" fill="#101727"/>
+<g font-family="Inter,Arial,'PingFang SC','Microsoft YaHei',sans-serif">
+<text x="40" y="55" font-size="28" fill="#fff" font-weight="600">One context. Two perspectives.</text><text x="40" y="88" font-size="17" fill="#b4bfd6" font-weight="400">One upstream Jev request · Choice and assessment returned together</text>
+<text x="40" y="146" font-size="18" fill="#a8b8ed" font-weight="400">HiRoute supplies</text><text x="792" y="146" font-size="18" fill="#a8b8ed" font-weight="400">HiRoute applies &amp; records</text><rect x="40" y="170" width="310" height="94" rx="15" fill="#182135" stroke="#33415d"/><text x="60" y="204" font-size="21" fill="#fff" font-weight="600">Current input</text><text x="60" y="237" font-size="16" fill="#b4bfd6" font-weight="400">Question, feedback or continuation</text><rect x="40" y="282" width="310" height="94" rx="15" fill="#182135" stroke="#33415d"/><text x="60" y="316" font-size="21" fill="#fff" font-weight="600">Execution history</text><text x="60" y="349" font-size="16" fill="#b4bfd6" font-weight="400">Answers · Tool activity · Branches</text><rect x="40" y="394" width="310" height="94" rx="15" fill="#182135" stroke="#33415d"/><text x="60" y="428" font-size="21" fill="#fff" font-weight="600">Allowed branches</text><text x="60" y="461" font-size="16" fill="#b4bfd6" font-weight="400">Defined by the active routing plan</text><path d="M350 217H380V329M350 441H380V329" fill="none" stroke="#8394cc" stroke-width="2"/><path d="M350 329H415" fill="none" stroke="#9ba9ee" stroke-width="2" marker-end="url(#arrow)"/><rect x="423" y="204" width="290" height="250" rx="15" fill="url(#hub)" stroke="#7582cb"/><text x="449" y="243" font-size="37" fill="#c0c9ff" font-weight="600">Jev</text><text x="449" y="275" font-size="18" fill="#b4bfd6" font-weight="400">Official Jev extension</text><text x="449" y="324" font-size="19" fill="#e9ecfb" font-weight="400">Understand the current task</text><text x="449" y="355" font-size="19" fill="#e9ecfb" font-weight="400">Consider prior performance</text><text x="449" y="416" font-size="18" fill="#c0c9ff" font-weight="400">Choice + optional Score</text><path d="M713 329H753V238H786" fill="none" stroke="#9ba9ee" stroke-width="2" marker-end="url(#arrow)"/><path d="M753 329V404H786" fill="none" stroke="#9ba9ee" stroke-width="2" marker-end="url(#arrow)"/><rect x="792" y="183" width="288" height="113" rx="15" fill="#222d4d" stroke="#7487d5"/><text x="812" y="222" font-size="23" fill="#d0d9ff" font-weight="600">Next branch</text><text x="812" y="260" font-size="17" fill="#b4bfd6" font-weight="400">For the work ahead</text><rect x="792" y="347" width="288" height="113" rx="15" fill="#2b2544" stroke="#8c78b6"/><text x="812" y="386" font-size="19" fill="#e0cef8" font-weight="600">Stage competence · optional</text><text x="812" y="424" font-size="16" fill="#c4b5d9" font-weight="400">Attributed to the previous stage</text><path d="M1080 238H1096V524H24V329H34" fill="none" stroke="#697aab" stroke-width="2" stroke-dasharray="6 6" marker-end="url(#arrow)"/><text x="265" y="557" font-size="18" fill="#b4bfd6" font-weight="400">Execution adds evidence for the next decision</text><text x="40" y="617" font-size="17" fill="#9aa8c6" font-weight="400">Assess the previous stage. Choose for the next one.</text></g></svg>
diff --git a/decision-extensions/assets/jev-decision-zh-CN.svg b/decision-extensions/assets/jev-decision-zh-CN.svg
new file mode 100644
index 00000000..4e6f25de
--- /dev/null
+++ b/decision-extensions/assets/jev-decision-zh-CN.svg
@@ -0,0 +1,7 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="1120" height="660" viewBox="0 0 1120 660" role="img" aria-labelledby="title desc">
+<title id="title">同一份证据,两种判断</title><desc id="desc">一次 Jev 上游请求 · 选择与评分同时返回. 评分回看此前阶段,分支选择面向下一阶段。</desc>
+<defs><marker id="arrow" markerWidth="8" markerHeight="8" refX="6" refY="4" orient="auto"><path d="M0 0L7 4L0 8" fill="none" stroke="#9ba9ee" stroke-width="1.5"/></marker><linearGradient id="hub" x2="1" y2="1"><stop stop-color="#303764"/><stop offset="1" stop-color="#222943"/></linearGradient></defs>
+<rect width="1120" height="660" rx="24" fill="#101727"/>
+<g font-family="Inter,Arial,'PingFang SC','Microsoft YaHei',sans-serif">
+<text x="40" y="55" font-size="28" fill="#fff" font-weight="600">同一份证据,两种判断</text><text x="40" y="88" font-size="17" fill="#b4bfd6" font-weight="400">一次 Jev 上游请求 · 选择与评分同时返回</text>
+<text x="40" y="146" font-size="18" fill="#a8b8ed" font-weight="400">HiRoute 提供</text><text x="792" y="146" font-size="18" fill="#a8b8ed" font-weight="400">HiRoute 执行与记录</text><rect x="40" y="170" width="310" height="94" rx="15" fill="#182135" stroke="#33415d"/><text x="60" y="204" font-size="21" fill="#fff" font-weight="600">当前输入</text><text x="60" y="237" font-size="16" fill="#b4bfd6" font-weight="400">问题、反馈或继续任务</text><rect x="40" y="282" width="310" height="94" rx="15" fill="#182135" stroke="#33415d"/><text x="60" y="316" font-size="21" fill="#fff" font-weight="600">执行历史</text><text x="60" y="349" font-size="16" fill="#b4bfd6" font-weight="400">回答 · 工具行为 · 分支</text><rect x="40" y="394" width="310" height="94" rx="15" fill="#182135" stroke="#33415d"/><text x="60" y="428" font-size="21" fill="#fff" font-weight="600">允许分支</text><text x="60" y="461" font-size="16" fill="#b4bfd6" font-weight="400">由当前路由计划限定</text><path d="M350 217H380V329M350 441H380V329" fill="none" stroke="#8394cc" stroke-width="2"/><path d="M350 329H415" fill="none" stroke="#9ba9ee" stroke-width="2" marker-end="url(#arrow)"/><rect x="423" y="204" width="290" height="250" rx="15" fill="url(#hub)" stroke="#7582cb"/><text x="449" y="243" font-size="37" fill="#c0c9ff" font-weight="600">Jev</text><text x="449" y="275" font-size="18" fill="#b4bfd6" font-weight="400">官方 Jev 扩展</text><text x="449" y="324" font-size="19" fill="#e9ecfb" font-weight="400">理解当前任务</text><text x="449" y="355" font-size="19" fill="#e9ecfb" font-weight="400">结合此前执行表现</text><text x="449" y="416" font-size="18" fill="#c0c9ff" font-weight="400">Choice + 可选 Score</text><path d="M713 329H753V238H786" fill="none" stroke="#9ba9ee" stroke-width="2" marker-end="url(#arrow)"/><path d="M753 329V404H786" fill="none" stroke="#9ba9ee" stroke-width="2" marker-end="url(#arrow)"/><rect x="792" y="183" width="288" height="113" rx="15" fill="#222d4d" stroke="#7487d5"/><text x="812" y="222" font-size="23" fill="#d0d9ff" font-weight="600">下一分支</text><text x="812" y="260" font-size="17" fill="#b4bfd6" font-weight="400">用于后续执行</text><rect x="792" y="347" width="288" height="113" rx="15" fill="#2b2544" stroke="#8c78b6"/><text x="812" y="386" font-size="19" fill="#e0cef8" font-weight="600">阶段胜任度 · 可选</text><text x="812" y="424" font-size="16" fill="#c4b5d9" font-weight="400">记录到此前模型阶段</text><path d="M1080 238H1096V524H24V329H34" fill="none" stroke="#697aab" stroke-width="2" stroke-dasharray="6 6" marker-end="url(#arrow)"/><text x="265" y="557" font-size="18" fill="#b4bfd6" font-weight="400">执行产生新证据,用于后续决策</text><text x="40" y="617" font-size="17" fill="#9aa8c6" font-weight="400">评分回看此前阶段,分支选择面向下一阶段。</text></g></svg>