omnilane 0.9.0 → 0.10.1

package/README.md

@@ -21,120 +21,72 @@ or any hosted model via OpenRouter — on the subscriptions you already pay for,
 
 ---
 
-## What's new in v0.9.0
+## 2026-07-25 update
 
-- **Five OpenAI-compatible direct-API vendors** — `deepseek`, `zai` (GLM),
-  `mistral`, `groq`, and `cerebras` join `openrouter` as CLI-free lanes (curl +
-  a `<VENDOR>_API_KEY`). A one-line `lib/common.sh` registry entry adds each;
-  see [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md).
-- **Fish shell completion** — `omnilane completion fish | source`.
+- Added `claude-opus-5` to default routing: first choice for `hard-judgment` and `taste-final`, plus a fallback for the hardest coding work.
+- Expanded `omnilane configure` across all 13 providers: 106 selectable model entries, complete live catalogs for Codex, Claude Code, Grok Build, and Antigravity, plus verified OpenRouter/OpenCode shortcuts. Custom model IDs remain available through `c`.
 
-## What's new in v0.8.3
-
-- **MCP server** — `omnilane mcp` starts a zero-dependency stdio MCP server,
-  so any MCP-capable host (Claude Code, Codex, Gemini CLI, Cursor, OpenCode…)
-  can discover and call omnilane without installing the skill: tools `route`,
-  `jobs_status`, `jobs_result`, and `list_lanes`. `route` defaults to
-  read-only advise mode; work mode requires an explicit workdir.
+---
+## 👋 New here?
 
-## What's new in v0.8.2
+You already drive an AI coding assistant — **Claude Code, Codex, Cursor, Gemini
+CLI**, and the like. Each one talks to a single AI model, and picking the best
+model for each task is left to you.
 
-- **`openrouter` vendor** — dispatch straight to the OpenRouter API with
-  nothing but `curl` and an `OPENROUTER_API_KEY`: hundreds of hosted models
-  become reachable from any omnilane install, no coding-agent CLI required.
-  Advise/consult only (it cannot edit files; work mode fails with guidance)
-  and the model slug is mandatory, e.g.
-  `dispatch.sh --vendor openrouter --model anthropic/claude-sonnet-5 consult "..."`.
-- **`deepseek`, `zai`, `mistral`, `groq`, `cerebras` vendors** — the same
-  CLI-free direct-API path as `openrouter`, for OpenAI-compatible providers:
-  DeepSeek, Z.ai GLM, Mistral, Groq, and Cerebras. Each needs only `curl` and
-  its `<VENDOR>_API_KEY`; advise/consult only. A one-line `lib/common.sh`
-  registry entry defines each endpoint, key env, and default model. See
-  [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md).
-- **`opencode` vendor** — headless dispatch through the OpenCode
-  multi-provider aggregator CLI (`opencode run`). Advise mode pins OpenCode's
-  built-in read-only `plan` agent; work mode uses `--auto`. Joins the default
-  `coding-overflow` chain as its last fallback.
+**omnilane picks for you.** For every piece of work it routes the task to the
+model that is best (and cheapest) for it — tough coding to a top coder, quick
+checks to a fast cheap model, long documents to a big-context model — all on the
+subscriptions and API keys you already pay for. Use the built-in defaults, or
+tweak one small file. Nothing new to babysit (it runs behind your existing tool),
+and `./install.sh --uninstall` removes it cleanly.
 
-## What's new in v0.8.1
+**[⬇ Jump to the 60-second start](#-60-second-start)**
 
-- **Claude Code plugin auto-loads the routing reminder** — the plugin now
-  ships a `SessionStart` hook (`hooks/hooks.json`) that injects the routing
-  reminder at session open (`startup|resume|clear`), so plugin installs get
-  the persistent reminder with no edit to `~/.claude/CLAUDE.md`. The
-  `install.sh` instruction-file reminder still covers the other CLIs.
+## What's new in v0.10.0
 
-## What's new in v0.8.0
+- **Gemini 3.6 Flash defaults** — the gemini candidates in `fast-agentic`,
+  `triage`, and `bulk-mechanical` (and the `Gemini Flash` alias) now run
+  Gemini 3.6 Flash (released 2026-07-21): fewer output tokens, a lower output
+  price, and the fastest output speed measured by Artificial Analysis.
+- **Evidence re-audit** — routing comments, model capability notes, and the
+  Gemini price table refreshed against official sources (2026-07-21/22).
 
-- **Two new dispatch vendors** — `kimi` (Moonshot Kimi Code CLI) and `qwen`
-  (Alibaba Qwen Code CLI) join the vendor set with the uniform runner
-  contract: advise stays read-only, work auto-approves, API-key env is
-  stripped so the CLIs use their own subscription logins, and empty output
-  is a loud failure. Pin them with `--vendor kimi|qwen`.
-- **coding-overflow grows a chain** — the quota relief valve now falls back
-  grok → kimi → qwen before `off`, so it works with any one of the three
-  vendors installed. Runners are contract-tested against fake binaries;
-  real-model reports welcome.
+## What's new in v0.9.1
 
-## What's new in v0.7.1
+- **Fix:** `configure set` no longer deletes hand-written comments from
+  `routing.local.yaml` — it rewrites only its own stamp header and the lane
+  being replaced.
 
-- **Routing refresh (2026-07 model data)** — hardest-coding now dispatches
-  GPT-5.6 Sol at **max** effort: Artificial Analysis Coding Agent Index v1.1
-  scores Sol (max) at 80, the current state of the art, retiring the older
-  xhigh-beats-max snapshot.
-- **Claude backups sharpened** — the Claude Opus 4.8 fallback on
-  hardest-coding and hard-judgment moves to **xhigh** effort, following
-  Anthropic's guidance to use extra effort for difficult tasks and
-  long-running work.
-
-## What's new in v0.7.0
-
-- **Preview any dispatch first** — `--dry-run` prints the fully resolved plan
-  (vendor, model, mode, timeouts, side-effect decision) with no provider call
-  and no job state.
-- **Automate with versioned JSON** — one `--json` envelope for `--list`,
-  `--explain`, `--validate`, and `jobs list|status|result|stats`, plus
-  read-only `jobs wait`, `jobs audit`, and an offline `omnilane release-audit`
-  gate with a deterministic manifest.
-- **Drive local jobs end to end** — `jobs tail` peeks at live output,
-  `jobs retry` re-dispatches a completed job fail-closed,
-  `prune --older-than` ages out old jobs, and `--help` covers every command.
-- **Install and complete safely** — `install.sh --check`/`--dry-run` report
-  drift without writing, `omnilane completion bash|zsh` ships safe tab
-  completion, and five macOS stock Bash 3.2 crashes are fixed.
+## What's new in v0.9.0
 
-## What's new in v0.6.0
+- **Five OpenAI-compatible direct-API vendors** — `deepseek`, `zai` (GLM),
+  `mistral`, `groq`, and `cerebras` join `openrouter` as CLI-free lanes (curl +
+  a `<VENDOR>_API_KEY`). A one-line `lib/common.sh` registry entry adds each;
+  see [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md).
+- **Fish shell completion** — `omnilane completion fish | source`.
 
-- **Explain and validate routes offline** — inspect every fallback candidate
-  with `--explain`, or lint the complete effective table with `--validate`,
-  without invoking a provider or creating job state.
-- **Inspect local health and outcomes** — bounded `jobs.sh stats` aggregates and
-  `omnilane doctor --json` make local automation observable without exposing
-  task or result bodies.
-- **Compare runs in Live Board** — pin one loaded job as a memory-only reference
-  and compare its model path and public result with the current selection.
-- **Keep lock recovery quiet** — transient owner-file read races no longer leak
-  misleading missing-file diagnostics.
+## ⚡ 60-second start
 
-## What's new in v0.5.1
+**The quick way — install from npm:**
 
-- **Use Codex work outside Git** — ordinary directories remain supported;
-  Omnilane never requires or runs `git init`.
-- **Stop non-Git hangs cleanly** — the resolved per-call watchdog becomes an
-  automatic process-group fuse when no whole-job timeout was configured, while
-  explicit timeout precedence and exit semantics remain intact.
-- **Trust the displayed version** — `VERSION` now drives `omnilane --version`
-  and both plugin manifests, with CI checking the changelog and all five READMEs.
+```bash
+npm i -g omnilane                                    # install the CLI
+omnilane route hardest-coding "fix the flaky auth token refresh"
+omnilane doctor                                      # see which AI CLIs / keys you have
+omnilane ui start                                    # optional: watch jobs live in your browser
+```
 
-## ⚡ 60-second start
+**Or clone the repo** (gets you the routing table and skill to customise):
 
 ```bash
 git clone https://github.com/Seraphim0916/omnilane && cd omnilane
 ./install.sh          # finds your CLIs, links the skill, speaks your language
 omnilane route hardest-coding "fix the flaky auth token refresh"
-omnilane ui start     # optional: watch jobs live in your browser
 ```
 
+> New to this? Run `omnilane doctor` first — it tells you which model CLIs and
+> API keys omnilane can already reach, so you know what will actually run.
+
 ## 🧭 How it works
 
 omnilane lets the main loop of **any** agentic CLI classify subtasks into
@@ -147,7 +99,7 @@ flowchart LR
     M["main loop<br/><i>any CLI you drive</i>"] --> T{{"routing.yaml<br/>one shared table"}}
     T -->|hardest-coding| C1["Codex — GPT-5.6 Sol"]
     T -->|bulk-mechanical| C2["Codex — GPT-5.6 Terra"]
-    T -->|taste-final| C3["Claude — Opus 4.8"]
+  T -->|taste-final| C3["Claude — Opus 5"]
     T -->|long-context| C4["Gemini — 3.1 Pro"]
     T -->|live-search| C5["Grok — 4.5"]
     T -->|"arbitrate (opt-in)"| C6["vote — 1-4 model panel"]
@@ -179,15 +131,15 @@ flowchart LR
 
 | Lane | First choice | Backup | When |
 |---|---|---|---|
-| 🔥 hardest-coding | GPT-5.6 Sol (max) | Claude Opus 4.8 (xhigh) | Hardest implementation, deep root-cause debug, correctness-critical edits |
+| 🔥 hardest-coding | GPT-5.6 Sol (max) | Claude Opus 5 (max) | Hardest implementation, deep root-cause debug, correctness-critical edits |
 | 🏗️ bulk-mechanical | GPT-5.6 Terra (max) | Claude Sonnet 5 (high) | Refactors, migrations, tests, review sweeps — mechanical endurance |
-| 🧹 triage | GPT-5.6 Luna (medium) | Gemini 3.5 Flash (Low) | High-volume scans, first-pass filtering |
-| ⚖️ hard-judgment | GPT-5.6 Sol (max) | Claude Opus 4.8 (xhigh) | Architecture arbitration, deep reasoning, second opinions |
-| ✒️ taste-final | Claude Opus 4.8 (high) | GPT-5.6 Sol (max) | User-facing prose, prompt/doc polish, style arbitration |
+| 🧹 triage | GPT-5.6 Luna (medium) | Gemini 3.6 Flash (Low) | High-volume scans, first-pass filtering |
+| ⚖️ hard-judgment | Claude Opus 5 (max) | GPT-5.6 Sol (max) | Architecture arbitration, deep reasoning, second opinions |
+| ✒️ taste-final | Claude Opus 5 (high) | GPT-5.6 Sol (max) | User-facing prose, prompt/doc polish, style arbitration |
 | 💬 consult | Explicit named vendor/model | — (no fallback) | Direct natural-language consultation; always keep `--vendor` |
-| 🎨 ui-draft | GPT-5.6 Sol (xhigh) | Claude Opus 4.8 (high) | UI drafts only WITH a design system / reference images |
-| 📚 long-context | Gemini 3.1 Pro (High) | Claude Opus 4.8 (high) | 1M-token synthesis — analysis only, never agentic loops |
-| ⚡ fast-agentic | Gemini 3.5 Flash (High) | GPT-5.6 Luna (high) | Fast multi-step agentic loops, multimodal checks |
+| 🎨 ui-draft | GPT-5.6 Sol (xhigh) | Claude Opus 5 (high) | UI drafts only WITH a design system / reference images |
+| 📚 long-context | Gemini 3.1 Pro (High) | Claude Opus 5 (high) | 1M-token synthesis; Pro is agentic-capable, while fast repeated loops prefer Flash |
+| ⚡ fast-agentic | Gemini 3.6 Flash (High) | GPT-5.6 Luna (high) | Fast multi-step agentic loops, multimodal checks |
 | 📡 live-search | Grok 4.5 | — (off) | Realtime X/web search and social context |
 | 🚰 coding-overflow | Grok 4.5 | Kimi K3 → Qwen3 Coder Plus → OpenCode | Codex-quota relief valve for mid-tier coding |
 | 🗳️ arbitrate | off (opt-in vote panel) | — | Built-in opinion panel for big calls — disabled by default; enable it in `routing.local.yaml`, one call per voter per round |
@@ -198,7 +150,9 @@ lane is such a chain; when nothing in it is installed the lane degrades to `off`
 
 > **Where is Claude Fable 5?** Deliberately not in the defaults: the top
 > Claude tier is usually the *main loop itself*, not a dispatched worker, and
-> it prices above Opus. It is offered in the configurator's model menu —
+> it prices above Opus. This is a cost / guardrail / main-loop policy choice,
+> not a capability verdict; Anthropic positions Fable 5 above Opus 5. It is
+> offered in the configurator's model menu —
 > route to it if you disagree (e.g. `taste-final: claude claude-fable-5 high`
 > in `routing.local.yaml`).
 
@@ -226,11 +180,11 @@ already are that model, so no second call) versus **dispatch**. Your harness's
 `omnilane` skill applies the right row automatically; this is the human view.
 
 - **Claude Code · Fable 5** — self-execute: hard-judgment, taste-final, the hardest correctness-critical fixes. Dispatch mechanical coding volume → Codex, long-context → Gemini, live-search → Grok.
-- **Claude Code · Opus 4.8** — self-execute: taste-final. Dispatch hard-judgment to Codex Sol (it out-scores Opus on raw intelligence), all coding to the Codex lanes, long-context → Gemini, live-search → Grok.
+- **Claude Code · Opus 5** — self-execute: hard-judgment and taste-final. Dispatch bulk coding to Codex lanes, long-context → Gemini, live-search → Grok.
 - **Codex · Sol** — self-execute: hardest-coding, hard-judgment, ui-draft. Dispatch taste-final → Claude, long-context → Gemini, live-search → Grok, bulk → Codex Terra.
 - **Codex · Terra** — self-execute: bulk-mechanical. Escalate the genuinely hardest pieces to Sol; dispatch taste → Claude, long-context → Gemini, live-search → Grok.
 - **Grok Build · Grok 4.5** — self-execute: live-search, coding-overflow (mid-tier coding). Dispatch everything hard to Codex/Claude/Gemini — and verify every API signature and cited fact first.
-- **Antigravity · Gemini** — self-execute: long-context (3.1 Pro) and fast-agentic (Flash). Dispatch coding/judgment/taste to Codex/Claude; live-search → Grok. Never take agentic tool-loop chains on 3.1 Pro.
+- **Antigravity · Gemini** — self-execute: long-context and context-heavy agentic work on 3.1 Pro, fast repeated loops on Flash. Dispatch hardest coding/judgment/taste to Codex/Claude; live-search → Grok.
 
 </details>
 
@@ -467,6 +421,110 @@ configurator and `routing.local.yaml` exist so you can disagree.
   a Git worktree, so the automatic fuse above bounds that case and cleans the
   supervised process group. Omnilane neither initializes nor requires a repository.
 
+## 📜 Release history
+
+<details>
+<summary>Older releases (v0.8.3 and earlier)</summary>
+
+## What's new in v0.8.3
+
+- **MCP server** — `omnilane mcp` starts a zero-dependency stdio MCP server,
+  so any MCP-capable host (Claude Code, Codex, Gemini CLI, Cursor, OpenCode…)
+  can discover and call omnilane without installing the skill: tools `route`,
+  `jobs_status`, `jobs_result`, and `list_lanes`. `route` defaults to
+  read-only advise mode; work mode requires an explicit workdir.
+
+## What's new in v0.8.2
+
+- **`openrouter` vendor** — dispatch straight to the OpenRouter API with
+  nothing but `curl` and an `OPENROUTER_API_KEY`: hundreds of hosted models
+  become reachable from any omnilane install, no coding-agent CLI required.
+  Advise/consult only (it cannot edit files; work mode fails with guidance)
+  and the model slug is mandatory, e.g.
+  `dispatch.sh --vendor openrouter --model anthropic/claude-sonnet-5 consult "..."`.
+- **`deepseek`, `zai`, `mistral`, `groq`, `cerebras` vendors** — the same
+  CLI-free direct-API path as `openrouter`, for OpenAI-compatible providers:
+  DeepSeek, Z.ai GLM, Mistral, Groq, and Cerebras. Each needs only `curl` and
+  its `<VENDOR>_API_KEY`; advise/consult only. A one-line `lib/common.sh`
+  registry entry defines each endpoint, key env, and default model. See
+  [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md).
+- **`opencode` vendor** — headless dispatch through the OpenCode
+  multi-provider aggregator CLI (`opencode run`). Advise mode pins OpenCode's
+  built-in read-only `plan` agent; work mode uses `--auto`. Joins the default
+  `coding-overflow` chain as its last fallback.
+
+## What's new in v0.8.1
+
+- **Claude Code plugin auto-loads the routing reminder** — the plugin now
+  ships a `SessionStart` hook (`hooks/hooks.json`) that injects the routing
+  reminder at session open (`startup|resume|clear`), so plugin installs get
+  the persistent reminder with no edit to `~/.claude/CLAUDE.md`. The
+  `install.sh` instruction-file reminder still covers the other CLIs.
+
+## What's new in v0.8.0
+
+- **Two new dispatch vendors** — `kimi` (Moonshot Kimi Code CLI) and `qwen`
+  (Alibaba Qwen Code CLI) join the vendor set with the uniform runner
+  contract: advise stays read-only, work auto-approves, API-key env is
+  stripped so the CLIs use their own subscription logins, and empty output
+  is a loud failure. Pin them with `--vendor kimi|qwen`.
+- **coding-overflow grows a chain** — the quota relief valve now falls back
+  grok → kimi → qwen before `off`, so it works with any one of the three
+  vendors installed. Runners are contract-tested against fake binaries;
+  real-model reports welcome.
+
+## What's new in v0.7.1
+
+- **Routing refresh (2026-07 model data)** — hardest-coding now dispatches
+  GPT-5.6 Sol at **max** effort: Artificial Analysis Coding Agent Index v1.1
+  scores Sol (max) at 80, the current state of the art, retiring the older
+  xhigh-beats-max snapshot.
+- **Claude backups sharpened** — the Claude Opus 4.8 fallback on
+  hardest-coding and hard-judgment moves to **xhigh** effort, following
+  Anthropic's guidance to use extra effort for difficult tasks and
+  long-running work.
+
+## What's new in v0.7.0
+
+- **Preview any dispatch first** — `--dry-run` prints the fully resolved plan
+  (vendor, model, mode, timeouts, side-effect decision) with no provider call
+  and no job state.
+- **Automate with versioned JSON** — one `--json` envelope for `--list`,
+  `--explain`, `--validate`, and `jobs list|status|result|stats`, plus
+  read-only `jobs wait`, `jobs audit`, and an offline `omnilane release-audit`
+  gate with a deterministic manifest.
+- **Drive local jobs end to end** — `jobs tail` peeks at live output,
+  `jobs retry` re-dispatches a completed job fail-closed,
+  `prune --older-than` ages out old jobs, and `--help` covers every command.
+- **Install and complete safely** — `install.sh --check`/`--dry-run` report
+  drift without writing, `omnilane completion bash|zsh` ships safe tab
+  completion, and five macOS stock Bash 3.2 crashes are fixed.
+
+## What's new in v0.6.0
+
+- **Explain and validate routes offline** — inspect every fallback candidate
+  with `--explain`, or lint the complete effective table with `--validate`,
+  without invoking a provider or creating job state.
+- **Inspect local health and outcomes** — bounded `jobs.sh stats` aggregates and
+  `omnilane doctor --json` make local automation observable without exposing
+  task or result bodies.
+- **Compare runs in Live Board** — pin one loaded job as a memory-only reference
+  and compare its model path and public result with the current selection.
+- **Keep lock recovery quiet** — transient owner-file read races no longer leak
+  misleading missing-file diagnostics.
+
+## What's new in v0.5.1
+
+- **Use Codex work outside Git** — ordinary directories remain supported;
+  Omnilane never requires or runs `git init`.
+- **Stop non-Git hangs cleanly** — the resolved per-call watchdog becomes an
+  automatic process-group fuse when no whole-job timeout was configured, while
+  explicit timeout precedence and exit semantics remain intact.
+- **Trust the displayed version** — `VERSION` now drives `omnilane --version`
+  and both plugin manifests, with CI checking the changelog and all five READMEs.
+
+</details>
+
 ## 🌱 Status
 
 omnilane now spans thirteen dispatch vendors — four harness natives (codex,

package/README.zh-CN.md

@@ -21,97 +21,64 @@
 
 ---
 
-## v0.9.0 新功能
-
-- **新增 5 个 OpenAI-compatible direct-API vendor** — `deepseek`、`zai`(GLM)、`mistral`、`groq`、`cerebras`,与 `openrouter` 同为免 CLI 通道(curl 加一把 `<VENDOR>_API_KEY`);`lib/common.sh` registry 一行即加一个。详见 [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md)。
-- **fish shell 补全** — `omnilane completion fish | source`。
-
-## v0.8.3 新功能
+## 2026-07-25 更新
 
-- **MCP server** — `omnilane mcp` 启动零依赖的 stdio MCP server,任何支持
-  MCP 的宿主(Claude Code、Codex、Gemini CLI、Cursor、OpenCode……)无需安装
-  skill 即可发现并调用 omnilane:提供 `route`、`jobs_status`、`jobs_result`、
-  `list_lanes` 四个工具。`route` 默认只读 advise 模式;work 模式必须明确
-  指定 workdir。
+- 默认路由加入 `claude-opus-5`：成为 `hard-judgment`、`taste-final` 第一顺位，也纳入最高难度编程任务的备选。
+- `omnilane configure` 已扩展全部 13 个提供商：共 106 个可选模型，完整收录 Codex、Claude Code、Grok Build、Antigravity 实时列表，并加入已验证的 OpenRouter／OpenCode 快捷项；仍可用 `c` 输入自定义模型 ID。
 
-## v0.8.2 新功能
+---
+## 👋 第一次用?
 
-- **`openrouter` vendor** — 只需 `curl` 加一个 `OPENROUTER_API_KEY`,
-  即可直连 OpenRouter API 派工:任何 omnilane 安装都能访问数百个
-  托管模型,无需再装任何代理 CLI。仅限 advise/consult(不能改文件,
-  work 模式会明确报错指路),模型 slug 必填,例如
-  `dispatch.sh --vendor openrouter --model anthropic/claude-sonnet-5 consult "..."`。
-- **`opencode` vendor** — 通过 OpenCode 多供应商聚合 CLI 无头派工
-  (`opencode run`)。advise 模式锁定内置只读 `plan` agent;work 模式
-  用 `--auto`。加入默认 `coding-overflow` 链作为最后回退。
+你已经在用某个 AI 写程式助手——**Claude Code、Codex、Cursor、Gemini
+CLI** 之类。每个一次只接一个 AI 模型,而「每件工作该用哪个模型最好」得你自己决定。
 
-## v0.8.1 新功能
+**omnilane 帮你决定。** 每一件工作,它会自动派给对那件事最强(也最省)的模型——硬派程式
+交给顶尖程式模型、随手的小检查交给又快又便宜的、长文件交给大脉络模型——全部用你本来就在
+付费的订阅与 API 金钥。用内建预设值就好,或改一个小设定档就能微调。不用另外顾一套东西
+(它躲在你现有工具背后跑),`./install.sh --uninstall` 可干净移除。
 
-- **Claude Code 插件开场自动载入路由提醒** — 插件新增 `SessionStart`
-  hook(`hooks/hooks.json`),在会话开始(`startup|resume|clear`)时自动
-  注入路由提醒,安装插件即生效,无需修改 `~/.claude/CLAUDE.md`。其他
-  CLI 仍使用 `install.sh` 的指令文件提醒。
+**[⬇ 直接跳到 60 秒上手](#-60-秒上手)**
 
-## v0.8.0 新功能
+## v0.10.0 新功能
 
-- **两个新派工 vendor** — `kimi`(Moonshot Kimi Code CLI)与 `qwen`
-  (Alibaba Qwen Code CLI)加入,沿用统一 runner 契约:advise 只读、
-  work 自动批准、剥除 API key 环境变量改用 CLI 自身订阅登录、空输出
-  视为失败。可用 `--vendor kimi|qwen` 直接指定。
-- **coding-overflow 新增备选链** — 额度溢流道改为 grok → kimi → qwen
-  再到 `off`,三家装任一家即可用。runner 以假可执行文件完成契约测试;
-  欢迎反馈真实模型实测结果。
+- **Gemini 3.6 Flash 默认路由**——`fast-agentic`、`triage`、`bulk-mechanical`
+  的 gemini 候选(与 `Gemini Flash` 别名)改用 2026-07-21 发布的 Gemini 3.6
+  Flash:输出 token 更少、输出单价更低、Artificial Analysis 实测输出速度第一。
+- **证据重审计**——路由注释、模型能力笔记与 Gemini 价格表对官方来源刷新
+  (2026-07-21/22)。
 
-## v0.7.1 新功能
+## v0.9.1 新功能
 
-- **路由表更新(2026-07 模型数据)** — hardest-coding 首选改为 GPT-5.6 Sol
-  **max** 档位:Artificial Analysis Coding Agent Index v1.1 测得 Sol (max)
-  80 分为当前最高,替换旧的「xhigh 胜 max」快照。
-- **Claude 备选升档** — hardest-coding 与 hard-judgment 的 Claude Opus 4.8
-  备选改为 **xhigh**,依 Anthropic 官方对困难任务与长时间工作的建议。
+- **修复**:`configure set` 不再删除 `routing.local.yaml` 中手写的注释——
+  只改写自身的戳记行与被替换的 lane。
 
-## v0.7.0 新功能
-
-- **先预览再派工** — `--dry-run` 打印完整解析后的派工计划(vendor、模型、
-  模式、超时、副作用判定),不调用模型、不创建作业状态。
-- **版本化 JSON 自动化** — `--list`/`--explain`/`--validate` 与
-  `jobs list|status|result|stats` 都提供 `--json` 信封;另有只读 `jobs wait`、
-  `jobs audit`,以及带可复现 manifest 的离线 `omnilane release-audit` 发布审计。
-- **本地作业一条龙** — `jobs tail` 查看实时输出、`jobs retry` 以 fail-closed
-  方式重派已完成作业、`prune --older-than` 按时间清理,`--help` 覆盖所有命令。
-- **安装与补全更安全** — `install.sh --check`/`--dry-run` 只读报告漂移,
-  `omnilane completion bash|zsh` 提供安全的 tab 补全,并修复五个 macOS 自带
-  Bash 3.2 崩溃。
+## v0.9.0 新功能
 
-## v0.6.0 新功能
+- **新增 5 个 OpenAI-compatible direct-API vendor** — `deepseek`、`zai`(GLM)、`mistral`、`groq`、`cerebras`,与 `openrouter` 同为免 CLI 通道(curl 加一把 `<VENDOR>_API_KEY`);`lib/common.sh` registry 一行即加一个。详见 [`docs/model-capabilities-2026-07.md`](docs/model-capabilities-2026-07.md)。
+- **fish shell 补全** — `omnilane completion fish | source`。
 
-- **离线理解并验证路由** — 使用 `--explain` 查看每个备用候选，或使用
-  `--validate` 检查完整生效路由表；都不会调用模型或创建作业状态。
-- **用机器可读数据观察本地状态** — `jobs.sh stats` 提供有界统计，
-  `omnilane doctor --json` 提供健康检查，同时不会泄漏任务或结果正文。
-- **在 Live Board 比较两条作业** — 将一条已加载作业固定为仅存在于内存中的
-  参考快照，并排比较模型路径与公开结果。
-- **让锁恢复更安静** — 所有者文件在检查与读取之间消失时，不再泄漏容易误判的
-  缺失文件诊断，同时保持 fail-closed。
+## ⚡ 60 秒上手
 
-## v0.5.1 新功能
+**最快的方式——用 npm 装:**
 
-- **在非 Git 目录使用 Codex work** — 普通文件夹仍完整支持；Omnilane 不要求，
-  也绝不会自动执行 `git init`。
-- **干净停止非 Git 卡死** — 未设置整体上限时，解析后的单次看门狗会自动成为
-  进程组保险丝，同时保留手动 timeout 的优先级和退出码语义。
-- **让版本显示可信** — `VERSION` 现在统一提供给 `omnilane --version` 和两份
-  plugin manifest，CI 会检查变更记录和五种语言 README 是否一致。
+```bash
+npm i -g omnilane                                    # 装 CLI
+omnilane route hardest-coding "修掉会间歇失败的 auth token 更新测试"
+omnilane doctor                                      # 看你手上有哪些 AI CLI / 金钥
+omnilane ui start                                    # 选配:在浏览器即时看派工
+```
 
-## ⚡ 60 秒上手
+**或 clone 整包**(拿到路由表与可自订的技能):
 
 ```bash
 git clone https://github.com/Seraphim0916/omnilane && cd omnilane
-./install.sh          # 检测你的 CLI、接好技能、说你的语言
-omnilane route hardest-coding "修掉间歇失败的 auth token 刷新测试"
-omnilane ui start     # 可选:在浏览器实时查看派发
+./install.sh          # 侦测你的 CLI、接好技能、说你的语言
+omnilane route hardest-coding "修掉会间歇失败的 auth token 更新测试"
 ```
 
+> 第一次用?先跑 `omnilane doctor`——它会告诉你 omnilane 现在能接到哪些模型 CLI 与
+> API 金钥,你就知道实际会跑什么。
+
 ## 🧭 工作原理
 
 omnilane 让**任何**一个 agentic CLI 的主循环把子任务分类到通道(lane),
@@ -123,7 +90,7 @@ flowchart LR
     M["主循环<br/><i>你在用的任一 CLI</i>"] --> T{{"routing.yaml<br/>一张共用路由表"}}
     T -->|hardest-coding| C1["Codex — GPT-5.6 Sol"]
     T -->|bulk-mechanical| C2["Codex — GPT-5.6 Terra"]
-    T -->|taste-final| C3["Claude — Opus 4.8"]
+    T -->|taste-final| C3["Claude — Opus 5"]
     T -->|long-context| C4["Gemini — 3.1 Pro"]
     T -->|live-search| C5["Grok — 4.5"]
     T -->|"arbitrate(可选)"| C6["vote — 1-4 模型评审团"]
@@ -152,15 +119,15 @@ flowchart LR
 
 | 通道 | 首选模型 | 备选模型 | 用途 |
 |---|---|---|---|
-| 🔥 hardest-coding | GPT-5.6 Sol (max) | Claude Opus 4.8 (xhigh) | 最难的实现、深度调试、正确性攸关的修改 |
+| 🔥 hardest-coding | GPT-5.6 Sol (max) | Claude Opus 5 (max) | 最难的实现、深度调试、正确性攸关的修改 |
 | 🏗️ bulk-mechanical | GPT-5.6 Terra (max) | Claude Sonnet 5 (high) | 重构、迁移、测试、大面积扫描——机械耐力活 |
-| 🧹 triage | GPT-5.6 Luna (medium) | Gemini 3.5 Flash (Low) | 高量初筛、第一轮过滤 |
-| ⚖️ hard-judgment | GPT-5.6 Sol (max) | Claude Opus 4.8 (xhigh) | 架构仲裁、深度推理、第二意见 |
-| ✒️ taste-final | Claude Opus 4.8 (high) | GPT-5.6 Sol (max) | 对外文字、prompt 与文档打磨、风格终审 |
+| 🧹 triage | GPT-5.6 Luna (medium) | Gemini 3.6 Flash (Low) | 高量初筛、第一轮过滤 |
+| ⚖️ hard-judgment | Claude Opus 5 (max) | GPT-5.6 Sol (max) | 架构仲裁、深度推理、第二意见 |
+| ✒️ taste-final | Claude Opus 5 (high) | GPT-5.6 Sol (max) | 对外文字、prompt 与文档打磨、风格终审 |
 | 💬 consult | 明确指定的厂商/模型 | —(不降级) | 自然语言直接咨询;必须保留 `--vendor` |
-| 🎨 ui-draft | GPT-5.6 Sol (xhigh) | Claude Opus 4.8 (high) | 有设计规范/参考图时的 UI 出稿;开放式视觉品味交给 taste-final |
-| 📚 long-context | Gemini 3.1 Pro (High) | Claude Opus 4.8 (high) | 百万 token 长文整合——仅限分析,不派 agentic 长链 |
-| ⚡ fast-agentic | Gemini 3.5 Flash (High) | GPT-5.6 Luna (high) | 快速多步骤 agentic 循环、多模态检查 |
+| 🎨 ui-draft | GPT-5.6 Sol (xhigh) | Claude Opus 5 (high) | 有设计规范/参考图时的 UI 出稿;开放式视觉品味交给 taste-final |
+| 📚 long-context | Gemini 3.1 Pro (High) | Claude Opus 5 (high) | 百万 token 长文整合;Pro 可跑 agentic 工作,高速重复循环仍优先 Flash |
+| ⚡ fast-agentic | Gemini 3.6 Flash (High) | GPT-5.6 Luna (high) | 快速多步骤 agentic 循环、多模态检查 |
 | 📡 live-search | Grok 4.5 | —(off) | 实时 X/网络搜索与社群脉络 |
 | 🚰 coding-overflow | Grok 4.5 | Kimi K3 → Qwen3 Coder Plus → OpenCode | Codex 额度吃紧时的中量级编码溢流道;事实性声明须另行查证 |
 | 🗳️ arbitrate | off(可选评审团) | — | 内置意见评审团,重大决定用——默认关闭,要用在 `routing.local.yaml` 打开;每评审每轮烧一次额度 |
@@ -168,7 +135,8 @@ flowchart LR
 **备选模型**是候选链的下一位——首选那家的厂商 CLI 没装时,派发就降到它。
 
 > **Claude Fable 5 去哪了?** 默认表刻意不放:Claude 顶级档通常就是*主循环本人*,
-> 不是被派发的工人,且定价高于 Opus。设置菜单的模型清单里有它——
+> 不是被派发的工人,且定价高于 Opus。这是成本/护栏/主循环策略,不代表能力
+> 不如 Opus;Anthropic 将 Fable 5 定位在 Opus 5 之上。设置菜单的模型清单里有它——
 > 不同意就自己路由过去(例如在 `routing.local.yaml` 写
 > `taste-final: claude claude-fable-5 high`)。
 
@@ -192,11 +160,11 @@ flowchart LR
 的 `omnilane` 技能会自动套对的那一行,这里是给人看的版本。
 
 - **Claude Code · Fable 5** — 自己做:hard-judgment、taste-final、最吃正确性的硬修。派出去:机械编码量 → Codex、长文 → Gemini、实时搜索 → Grok。
-- **Claude Code · Opus 4.8** — 自己做:taste-final。hard-judgment 派给 Codex Sol(智力分高于 Opus)、所有编码走 Codex 通道、长文 → Gemini、实时搜索 → Grok。
+- **Claude Code · Opus 5** — 自己做：hard-judgment、taste-final。大量编码走 Codex 通道、长文 → Gemini、实时搜索 → Grok。
 - **Codex · Sol** — 自己做:hardest-coding、hard-judgment、ui-draft。派出去:taste-final → Claude、长文 → Gemini、实时搜索 → Grok、粗活 → Codex Terra。
 - **Codex · Terra** — 自己做:bulk-mechanical。真正最硬的往上升给 Sol;taste → Claude、长文 → Gemini、实时搜索 → Grok。
 - **Grok Build · Grok 4.5** — 自己做:live-search、coding-overflow(中量级编码)。所有硬活派给 Codex/Claude/Gemini——先验每个 API 签名与引用事实。
-- **Antigravity · Gemini** — 自己做:long-context(3.1 Pro)、fast-agentic(Flash)。编码/判断/文字派给 Codex/Claude;实时搜索 → Grok。3.1 Pro 绝不接 agentic 工具长链。
+- **Antigravity · Gemini** — 自己做:3.1 Pro 的长文与重上下文 agentic 工作、Flash 的高速重复循环。最难的编码/判断/文字派给 Codex/Claude;实时搜索 → Grok。
 
 </details>
 
@@ -394,6 +362,90 @@ configure.sh set|get|unset|list|diff LANE [SPEC]    # 非交互编辑/查看 rou
   外卡住，因此上面的自动保险丝会限制这个场景并清理受监工的进程组。Omnilane
   不会自动执行 `git init`，也不要求用户创建仓库。
 
+## 📜 版本历程
+
+<details>
+<summary>旧版本(v0.8.3 以前)</summary>
+
+## v0.8.3 新功能
+
+- **MCP server** — `omnilane mcp` 启动零依赖的 stdio MCP server,任何支持
+  MCP 的宿主(Claude Code、Codex、Gemini CLI、Cursor、OpenCode……)无需安装
+  skill 即可发现并调用 omnilane:提供 `route`、`jobs_status`、`jobs_result`、
+  `list_lanes` 四个工具。`route` 默认只读 advise 模式;work 模式必须明确
+  指定 workdir。
+
+## v0.8.2 新功能
+
+- **`openrouter` vendor** — 只需 `curl` 加一个 `OPENROUTER_API_KEY`,
+  即可直连 OpenRouter API 派工:任何 omnilane 安装都能访问数百个
+  托管模型,无需再装任何代理 CLI。仅限 advise/consult(不能改文件,
+  work 模式会明确报错指路),模型 slug 必填,例如
+  `dispatch.sh --vendor openrouter --model anthropic/claude-sonnet-5 consult "..."`。
+- **`opencode` vendor** — 通过 OpenCode 多供应商聚合 CLI 无头派工
+  (`opencode run`)。advise 模式锁定内置只读 `plan` agent;work 模式
+  用 `--auto`。加入默认 `coding-overflow` 链作为最后回退。
+
+## v0.8.1 新功能
+
+- **Claude Code 插件开场自动载入路由提醒** — 插件新增 `SessionStart`
+  hook(`hooks/hooks.json`),在会话开始(`startup|resume|clear`)时自动
+  注入路由提醒,安装插件即生效,无需修改 `~/.claude/CLAUDE.md`。其他
+  CLI 仍使用 `install.sh` 的指令文件提醒。
+
+## v0.8.0 新功能
+
+- **两个新派工 vendor** — `kimi`(Moonshot Kimi Code CLI)与 `qwen`
+  (Alibaba Qwen Code CLI)加入,沿用统一 runner 契约:advise 只读、
+  work 自动批准、剥除 API key 环境变量改用 CLI 自身订阅登录、空输出
+  视为失败。可用 `--vendor kimi|qwen` 直接指定。
+- **coding-overflow 新增备选链** — 额度溢流道改为 grok → kimi → qwen
+  再到 `off`,三家装任一家即可用。runner 以假可执行文件完成契约测试;
+  欢迎反馈真实模型实测结果。
+
+## v0.7.1 新功能
+
+- **路由表更新(2026-07 模型数据)** — hardest-coding 首选改为 GPT-5.6 Sol
+  **max** 档位:Artificial Analysis Coding Agent Index v1.1 测得 Sol (max)
+  80 分为当前最高,替换旧的「xhigh 胜 max」快照。
+- **Claude 备选升档** — hardest-coding 与 hard-judgment 的 Claude Opus 4.8
+  备选改为 **xhigh**,依 Anthropic 官方对困难任务与长时间工作的建议。
+
+## v0.7.0 新功能
+
+- **先预览再派工** — `--dry-run` 打印完整解析后的派工计划(vendor、模型、
+  模式、超时、副作用判定),不调用模型、不创建作业状态。
+- **版本化 JSON 自动化** — `--list`/`--explain`/`--validate` 与
+  `jobs list|status|result|stats` 都提供 `--json` 信封;另有只读 `jobs wait`、
+  `jobs audit`,以及带可复现 manifest 的离线 `omnilane release-audit` 发布审计。
+- **本地作业一条龙** — `jobs tail` 查看实时输出、`jobs retry` 以 fail-closed
+  方式重派已完成作业、`prune --older-than` 按时间清理,`--help` 覆盖所有命令。
+- **安装与补全更安全** — `install.sh --check`/`--dry-run` 只读报告漂移,
+  `omnilane completion bash|zsh` 提供安全的 tab 补全,并修复五个 macOS 自带
+  Bash 3.2 崩溃。
+
+## v0.6.0 新功能
+
+- **离线理解并验证路由** — 使用 `--explain` 查看每个备用候选，或使用
+  `--validate` 检查完整生效路由表；都不会调用模型或创建作业状态。
+- **用机器可读数据观察本地状态** — `jobs.sh stats` 提供有界统计，
+  `omnilane doctor --json` 提供健康检查，同时不会泄漏任务或结果正文。
+- **在 Live Board 比较两条作业** — 将一条已加载作业固定为仅存在于内存中的
+  参考快照，并排比较模型路径与公开结果。
+- **让锁恢复更安静** — 所有者文件在检查与读取之间消失时，不再泄漏容易误判的
+  缺失文件诊断，同时保持 fail-closed。
+
+## v0.5.1 新功能
+
+- **在非 Git 目录使用 Codex work** — 普通文件夹仍完整支持；Omnilane 不要求，
+  也绝不会自动执行 `git init`。
+- **干净停止非 Git 卡死** — 未设置整体上限时，解析后的单次看门狗会自动成为
+  进程组保险丝，同时保留手动 timeout 的优先级和退出码语义。
+- **让版本显示可信** — `VERSION` 现在统一提供给 `omnilane --version` 和两份
+  plugin manifest，CI 会检查变更记录和五种语言 README 是否一致。
+
+</details>
+
 ## 🌱 状态
 
 omnilane 现有 13 个派工 vendor——4 个框架原生(codex、claude、grok、gemini)、3 个聚合/溢流 CLI(kimi、qwen、opencode),加上 6 个免 CLI 的 OpenAI-compatible direct-API vendor(openrouter、deepseek、zai、mistral、groq、cerebras)——全部走统一 runner 契约并附 contract 测试,另有 Claude Code `SessionStart` 自动提醒与 MCP stdio server 介面(`omnilane mcp`)。direct-API 与聚合 runner 皆以假可执行档做过契约测试;欢迎反馈真实模型使用经验。Grok/Antigravity 命令壳行为仍可能随 CLI 版本变动。欢迎提交 issue 与 PR。