feat(dsh): remote managed-agent as on-demand tool + bl managed-agent run

Rework the managed-agent integration so a dsh user can, in plain language, have a Bailian cloud agent created and run a task — no hand-written agents.yaml, no prior apply. New `bl managed-agent run --prompt <task> [--instructions] [--model] [--agent]`: one step that idempotently materializes a cloud agent + its environment, then opens a session and streams the result. It mirrors the OpenAgentPack webui backend's ensure+run recipe (resolveProjectConfigFrom Object → syncAgentResourcesWithStateBackend → readProjectRuntime + startSessionRun) from an in-memory config, reusing the existing credential spine in _engine/credentials.ts. State persists under the bl config dir (~/.bailian/managed-agent/<agent>/), never the user's cwd, so repeat runs with the same --agent reuse the materialized agent. Unlike apply it provisions without --yes, since running is the intent. dsh side: replace the SubagentProvider with a plain tool `bailian_run_remote_task` (packages/dsh/src/tool-managed-agent). The subagent seam did not fit: in the web profile every tool-subagent row is disabled in the host plane (delegation lives in agent presets), a provider fixes one agent identity in config, and the default numeric maxDepth would fail-mount a no-depthLimit provider. As a tool the model calls it directly and fills `instructions` from the user's intent, so the remote agent's role is defined per task. Enabled by default — it creates nothing at load, only on invocation. LLM row: configure the base bundle's existing llm-pi-ai row instead of mounting a second pi-ai instance (a second instance re-declares pi-ai's global configurable-provider catalog and fails boot on a duplicate amazon-bedrock). TokenPlan reads a dedicated BAILIAN_TOKENPLAN_API_KEY, not DASHSCOPE_API_KEY: TokenPlan (sk-sp-) and pay-as-you-go (sk-ws-) keys 401 each other's endpoints, so sharing one var would silently break whichever plugin lost. Note: the ensure+run happy path could not be verified end-to-end on the available account — agentstudio returns 404 there, and the existing `managed-agent apply` 404s identically against the same endpoint/key, so the failure is account/service provisioning, not this change. Command wiring, dry-run, config assembly, credential injection and URL construction were all verified. Co-Authored-By: Claude <noreply@anthropic.com>
feat(dsh): add bailian-cli-dsh plugin bundle for DeepSeek Harness
2026-09-14 19:49:23 +08:00 · 2026-08-15 10:29:24 +08:00 · 2026-08-15 01:47:18 +08:00 · 2026-08-07 18:02:08 +08:00 · 2026-08-07 17:42:31 +08:00 · 2026-08-07 17:26:21 +08:00
206 changed files with 13457 additions and 1700 deletions
@@ -18,7 +18,7 @@ on:
          - channel
          - stable
      channel:
-        description: "dist-tag (channel mode only, e.g. mcp/plugin/advisor)"
+        description: "Required when mode=channel. npm dist-tag only (lowercase, digits, dashes), e.g. mcp / plugin / sync-release. bailian-cli binary CDN always overwrites sync-release.json; knowledge-studio-cli is npm-only."
        required: false
        type: string

@@ -29,11 +29,11 @@ concurrency:
 jobs:
  publish-stable:
    if: inputs.mode == 'stable'
-    name: publish stable (${{ inputs.package }}) to npm + tag
+    name: publish stable (${{ inputs.package }}) to npm + binary + tag
    runs-on: ubuntu-latest
    environment: production # Required Reviewers gate
    permissions:
-      contents: write # push lightweight tag to origin
+      contents: write # push tag + create GitHub Release with binary assets
      id-token: write # OIDC for npm Trusted Publishing + provenance
    steps:
      - uses: actions/checkout@v6
@@ -55,19 +55,47 @@ jobs:
            | sudo tar -xz -C /usr/local/bin gitleaks
          gitleaks version

+      - name: Ensure zip (per-platform binary archives)
+        run: sudo apt-get update && sudo apt-get install -y zip
+
      - run: pnpm install --frozen-lockfile

+      # Binary compile uses `bun build --compile` CLI (not Bun.build API).
+      # Keep this pin in sync with any local smoke tests of binary-compile.mjs.
+      - uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: "1.2.19"
+
      - name: publish-stable
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          # OSS release channel runs fully in CI: upload + reconcile + manifest.json.
+          # All values come from repo Settings → Secrets — no OSS defaults live in
+          # code. Leave AK/SK unset to skip the OSS channel; once enabled,
+          # bucket/region/prefix are required.
+          BAILIAN_OSS_AK: ${{ secrets.BAILIAN_OSS_AK }}
+          BAILIAN_OSS_SK: ${{ secrets.BAILIAN_OSS_SK }}
+          BAILIAN_OSS_BUCKET: ${{ secrets.BAILIAN_OSS_BUCKET }}
+          BAILIAN_OSS_REGION: ${{ secrets.BAILIAN_OSS_REGION }}
+          BAILIAN_OSS_ENDPOINT: ${{ secrets.BAILIAN_OSS_ENDPOINT }}
+          BAILIAN_RELEASE_PREFIX: ${{ secrets.BAILIAN_RELEASE_PREFIX }}
+          BAILIAN_STATIC_PREFIX: ${{ secrets.BAILIAN_STATIC_PREFIX }}
        run: node tools/release/publish-stable.mjs ${{ inputs.package == 'knowledge-studio-cli' && '--knowledge' || '' }}

  publish-channel:
    if: inputs.mode == 'channel'
-    name: publish channel (${{ inputs.package }}) to npm
+    name: publish channel (${{ inputs.package }}) to npm + binary
    runs-on: ubuntu-latest
    permissions:
-      contents: read # no tag, no Release; just publish
+      contents: write # create prerelease GitHub Release with binary assets
      id-token: write # OIDC for npm Trusted Publishing + provenance
    steps:
+      - name: Require channel input
+        if: ${{ inputs.channel == '' }}
+        run: |
+          echo "::error::mode=channel requires the workflow input \"channel\" (npm dist-tag, e.g. mcp / plugin / sync-release). Leave mode=stable if you do not need a dist-tag."
+          exit 1
+
      - uses: actions/checkout@v6

      - uses: pnpm/action-setup@v6
@@ -87,7 +115,26 @@ jobs:
            | sudo tar -xz -C /usr/local/bin gitleaks
          gitleaks version

+      - name: Ensure zip (per-platform binary archives)
+        run: sudo apt-get update && sudo apt-get install -y zip
+
      - run: pnpm install --frozen-lockfile

+      # Binary compile uses `bun build --compile` CLI (not Bun.build API).
+      # Keep this pin in sync with any local smoke tests of binary-compile.mjs.
+      - uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: "1.2.19"
+
      - name: publish-channel
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          # OSS release channel — same Settings-injected values as stable.
+          BAILIAN_OSS_AK: ${{ secrets.BAILIAN_OSS_AK }}
+          BAILIAN_OSS_SK: ${{ secrets.BAILIAN_OSS_SK }}
+          BAILIAN_OSS_BUCKET: ${{ secrets.BAILIAN_OSS_BUCKET }}
+          BAILIAN_OSS_REGION: ${{ secrets.BAILIAN_OSS_REGION }}
+          BAILIAN_OSS_ENDPOINT: ${{ secrets.BAILIAN_OSS_ENDPOINT }}
+          BAILIAN_RELEASE_PREFIX: ${{ secrets.BAILIAN_RELEASE_PREFIX }}
+          BAILIAN_STATIC_PREFIX: ${{ secrets.BAILIAN_STATIC_PREFIX }}
        run: node tools/release/publish-channel.mjs ${{ inputs.package == 'knowledge-studio-cli' && '--knowledge' || '' }} --channel "${{ inputs.channel }}"
@@ -10,6 +10,7 @@ lerna-debug.log*
 # Dependencies & build output
 node_modules
 dist
+dist-bin
 dist-ssr
 tools/generated
 .node-version
@@ -36,7 +37,9 @@ tools/generated
 .claude/settings.local.json
 .claude/scheduled_tasks.lock
 .cursor/
+.qoder/
 .qwen/
+.qoder
 .playwright-mcp/
 .pnpm-store/

@@ -46,3 +49,9 @@ packages/cli/scene/**/outputs/

 # Environment variables (sensitive data)
 .env
+
+# Local scratch / plan drafts (never commit)
+.scratch/
+
+# pnpm pack output
+*.tgz
@@ -5,6 +5,15 @@ set -eu
 pnpm run sync:skill-assets

 # Stage generator output so it is included in this commit.
-git add skills/bailian-cli/reference skills/bailian-cli/SKILL.md
+git add \
+  skills/bailian-protocol/SKILL.md \
+  skills/bailian-cli/SKILL.md \
+  skills/bailian-cli/reference \
+  skills/bailian-gen/SKILL.md \
+  skills/bailian-gen/reference \
+  skills/bailian-finetune/SKILL.md \
+  skills/bailian-finetune/reference \
+  skills/bailian-managed-agent/SKILL.md \
+  skills/bailian-managed-agent/reference

 vp staged
@@ -35,7 +35,7 @@ packages/core/src/auth/            # apiKey / console credential 解析与落盘
 packages/core/src/client/          # HTTP client / endpoints / console gateway
 ```

-Skill / 命令手册随 `skills/bailian-cli/` 经 `npx skills add modelstudioai/cli` 安装。`tools/generate-reference.ts` 从 **`packages/cli/src/commands.ts`** 生成 `skills/bailian-cli/reference/`(纳入 git);`tools/sync-skill-metadata.ts` 从 `packages/cli/package.json` 同步 `skills/bailian-cli/SKILL.md` 的 `metadata.version`。两者由根脚本 `pnpm run sync:skill-assets` 和 `.vite-hooks/pre-commit` 执行。
+Skill / 命令手册随 `skills/bailian-*/` 经 `npx skills add modelstudioai/cli --all -g` 安装（整包装齐，含共享协议 `bailian-protocol`）。业务 skill（`bailian-cli` / `bailian-gen` / `bailian-finetune` / `bailian-managed-agent`）执行前读 `skills/bailian-protocol/`；不要依赖 frontmatter `companions`（安装器不强制）。`tools/generate-reference.ts` 从 **`packages/cli/src/commands.ts`** 按一级命令归属表分流写入各 `skills/<skill>/reference/`(纳入 git);`tools/sync-skill-metadata.ts` 从 `packages/cli/package.json` 同步各 `skills/*/SKILL.md` 的 `metadata.version`。两者由根脚本 `pnpm run sync:skill-assets` 和 `.vite-hooks/pre-commit` 执行。hub `bailian-cli` 的路由表不复述领域命令明细；SKILL 文案 / 安装约定 / hand-off 见 [docs/agents/skill-change.md](docs/agents/skill-change.md)。

 约定:

@@ -48,31 +48,33 @@ Skill / 命令手册随 `skills/bailian-cli/` 经 `npx skills add modelstudioai/
 非代码资产:

 - `tools/release/` — 发版自动化（CI 驱动,见 `.github/workflows/publish.yml`）
- `tools/generate-reference.ts` — 从 `packages/cli/src/commands.ts` 生成 `skills/bailian-cli/reference/`
- `tools/sync-skill-metadata.ts` — 同步 `skills/bailian-cli/SKILL.md` 的 `metadata.version`
+- `tools/generate-reference.ts` — 从 `packages/cli/src/commands.ts` 按归属表生成各 `skills/<skill>/reference/`
+- `tools/sync-skill-metadata.ts` — 同步各 `skills/*/SKILL.md` 的 `metadata.version`（含 `bailian-protocol`）
 - `README.md` / `README.zh.md` — npm 和 GitHub 主页

 ## 业务场景索引

 按当前任务从下表挑一条进入对应文档:

-| 场景           | 何时进入                                     | 详见                                                                         |
-| -------------- | -------------------------------------------- | ---------------------------------------------------------------------------- |
-| 命令增删改     | 增加 / 删除 / 重命名 `bl xxx` 或入口命令路径 | [docs/agents/command-add-remove.md](docs/agents/command-add-remove.md)       |
-| E2E 测试维护   | 新增/改命令或 e2e 用例、补 help/缺参/dry-run | [docs/agents/cli-e2e-tests.md](docs/agents/cli-e2e-tests.md)                 |
-| 批量压测       | 改/跑多能力并发压测、`test:stress`、fixtures | [docs/agents/stress-batch-tests.md](docs/agents/stress-batch-tests.md)       |
-| 选项变更       | 给已有命令加 `--flag` 或改默认值             | [docs/agents/command-flag-change.md](docs/agents/command-flag-change.md)     |
-| 模型上下架     | 增加新模型 / 改默认模型 / 废弃旧模型         | [docs/agents/model-add-remove.md](docs/agents/model-add-remove.md)           |
-| 错误文案变更   | 改 `BailianError` 的 message 或 hint         | [docs/agents/error-hint-change.md](docs/agents/error-hint-change.md)         |
-| URL / 渠道变更 | 控制台域名 / 文档站 / 追踪参数               | [docs/agents/url-change.md](docs/agents/url-change.md)                       |
-| 鉴权扩展       | 加 OAuth / SSO / 换 token 来源               | [docs/agents/auth-change.md](docs/agents/auth-change.md)                     |
-| 配置项扩展     | 新 env var 或 `~/.bailian/config.json` 字段  | [docs/agents/config-add.md](docs/agents/config-add.md)                       |
-| Profile / 激活 | 改命名 Profile、预设或 `active_config`       | [docs/agents/config-profile-change.md](docs/agents/config-profile-change.md) |
-| 安装文档       | 改安装、鉴权、验证流程或线上 install 页面    | [docs/agents/install-doc-change.md](docs/agents/install-doc-change.md)       |
-| 发布           | channel / stable 发布到 npm（CI 驱动）       | [docs/agents/publish.md](docs/agents/publish.md)                             |
-| Change Log     | 发版说明 / 历史版本说明                      | [docs/agents/changelog-write.md](docs/agents/changelog-write.md)             |
-| 工具链调整     | lint 规则 / 构建配置 / 依赖升级              | [docs/agents/lint-toolchain.md](docs/agents/lint-toolchain.md)               |
-| Command Pack   | 扩展包 / 白名单 / plugin 管理命令            | [docs/agents/command-pack.md](docs/agents/command-pack.md)                   |
+| 场景              | 何时进入                                        | 详见                                                                         |
+| ----------------- | ----------------------------------------------- | ---------------------------------------------------------------------------- |
+| 命令增删改        | 增加 / 删除 / 重命名 `bl xxx` 或入口命令路径    | [docs/agents/command-add-remove.md](docs/agents/command-add-remove.md)       |
+| E2E 测试维护      | 新增/改命令或 e2e 用例、补 help/缺参/dry-run    | [docs/agents/cli-e2e-tests.md](docs/agents/cli-e2e-tests.md)                 |
+| 批量压测          | 改/跑多能力并发压测、`test:stress`、fixtures    | [docs/agents/stress-batch-tests.md](docs/agents/stress-batch-tests.md)       |
+| 选项变更          | 给已有命令加 `--flag` 或改默认值                | [docs/agents/command-flag-change.md](docs/agents/command-flag-change.md)     |
+| 模型上下架        | 增加新模型 / 改默认模型 / 废弃旧模型            | [docs/agents/model-add-remove.md](docs/agents/model-add-remove.md)           |
+| Skill 文案 / 路由 | 改 SKILL 路由、安装约定、hand-off、hub/领域边界 | [docs/agents/skill-change.md](docs/agents/skill-change.md)                   |
+| 错误文案变更      | 改 `BailianError` 的 message 或 hint            | [docs/agents/error-hint-change.md](docs/agents/error-hint-change.md)         |
+| URL / 渠道变更    | 控制台域名 / 文档站 / 追踪参数                  | [docs/agents/url-change.md](docs/agents/url-change.md)                       |
+| 埋点变更          | 改 AEM 命令事件、后端渠道 header、User-Agent    | [docs/agents/telemetry-change.md](docs/agents/telemetry-change.md)           |
+| 鉴权扩展          | 加 OAuth / SSO / 换 token 来源                  | [docs/agents/auth-change.md](docs/agents/auth-change.md)                     |
+| 配置项扩展        | 新 env var 或 `~/.bailian/config.json` 字段     | [docs/agents/config-add.md](docs/agents/config-add.md)                       |
+| Profile / 激活    | 改命名 Profile、预设或 `active_config`          | [docs/agents/config-profile-change.md](docs/agents/config-profile-change.md) |
+| 安装文档          | 改安装、鉴权、验证流程或线上 install 页面       | [docs/agents/install-doc-change.md](docs/agents/install-doc-change.md)       |
+| 发布              | channel / stable 发布到 npm（CI 驱动）          | [docs/agents/publish.md](docs/agents/publish.md)                             |
+| Change Log        | 发版说明 / 历史版本说明                         | [docs/agents/changelog-write.md](docs/agents/changelog-write.md)             |
+| 工具链调整        | lint 规则 / 构建配置 / 依赖升级                 | [docs/agents/lint-toolchain.md](docs/agents/lint-toolchain.md)               |
+| Command Pack      | 扩展包 / 白名单 / plugin 管理命令               | [docs/agents/command-pack.md](docs/agents/command-pack.md)                   |

 如果当前任务无法对应任何场景,先按经验完成,然后**回来评估这是不是一类新场景** —— 是就新增 `docs/agents/<scenario>.md`,把清单沉淀下来。

@@ -6,6 +6,34 @@ The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and

 [中文版](CHANGELOG.zh.md) · [README](README.md) · [Contributing](CONTRIBUTING.md)

+## [1.14.1] - 2026-08-05
+
+### Added
+
+- **Focused Bailian Skills** — `npx skills add modelstudioai/cli --all -g` now installs dedicated skills for media generation, fine-tuning, Managed Agent, and shared execution rules, improving task routing while reducing irrelevant context.
+
+### Changed
+
+- **Default image model upgraded to Qwen-Image 3.0** — image generation, image editing, pipelines, the config UI, and related documentation now default to `qwen-image-3.0` for API Key users.
+- **Broader coding-agent compatibility** — Skill installation and updates now detect more coding agents, preserve existing installation links, and automatically backfill skills into newly detected agents.
+
+## [1.14.0] - 2026-08-04
+
+### Added
+
+- **Standalone installation without Node.js** — binary packages are available for macOS on Apple Silicon and Intel, Linux x64, and Windows x64; npm installation remains supported.
+- **Exact-version updates** — binary and npm installations can use `bl update --to <version>` to update or switch to a specified version.
+
+### Changed
+
+- **Binary self-updates** — binary installations now check and download updates through a dedicated release channel. `bl update` no longer replaces the running executable, and the next invocation automatically uses the new version.
+
+## [1.13.1] - 2026-08-03
+
+### Changed
+
+- **Default text model upgraded to Qwen3.8-Max** — `bl text chat`, pipelines, API key validation, the config UI, and Managed Agent init templates now default to `qwen3.8-max`; Token Plan also moves from the preview model to the stable release.
+
 ## [1.13.0] - 2026-07-30

 ### Added
@@ -6,6 +6,34 @@

 [English](CHANGELOG.md) · [README](README.zh.md) · [参与贡献](CONTRIBUTING.zh.md)

+## [1.14.1] - 2026-08-05
+
+### 新增
+
+- **百炼 Skill 按领域拆分** —— 通过 `npx skills add modelstudioai/cli --all -g` 可统一安装图片与视频生成、模型微调、Managed Agent 和共享执行协议等专用 Skill，提升任务路由准确性并减少无关上下文。
+
+### 变更
+
+- **默认图片模型升级至 Qwen-Image 3.0** —— 普通 API Key 用户的图片生成、图片编辑、Pipeline、配置 UI 和相关文档现在默认使用 `qwen-image-3.0`。
+- **扩展 Coding Agent 兼容范围** —— Skill 安装与更新现在能够识别更多 Coding Agent，保留已有安装链接，并自动将 Skill 补充到新识别的 Agent。
+
+## [1.14.0] - 2026-08-04
+
+### 新增
+
+- **免 Node.js 的二进制安装** — 支持 macOS Apple Silicon / Intel、Linux x64 和 Windows x64；npm 安装方式继续保留。
+- **指定版本更新** — 二进制和 npm 安装均可通过 `bl update --to <version>` 更新或切换到指定版本。
+
+### 变更
+
+- **二进制自更新** — 二进制安装现在通过独立的发布通道检查和下载更新；执行 `bl update` 时不会覆盖正在运行的程序，下次运行自动使用新版本。
+
+## [1.13.1] - 2026-08-03
+
+### 变更
+
+- **默认文本模型升级至 Qwen3.8-Max** — `bl text chat`、Pipeline、API Key 登录校验、配置 UI 和 Managed Agent 初始化模板现在默认使用 `qwen3.8-max`；Token Plan 也由预览版切换至正式版。
+
 ## [1.13.0] - 2026-07-30

 ### 新增
@@ -1,58 +1,71 @@
 # 阿里云百炼CLI 安装说明（供 AI Agent 阅读）

-本文档面向 **AI Agent**：按下列步骤为用户安装并校验 **阿里云百炼CLI**（npm 包名 `bailian-cli`，命令 `bl` / `bailian`）。用户通常通过 **npm 全局安装** 使用，**无需**访问本仓库源码。不要臆造版本号或路径；以用户环境为准。
+本文档面向 **AI Agent**：按下列步骤为用户安装并校验 **阿里云百炼CLI**（命令 `bl` / `bailian`）。**优先**使用二进制一键安装（无需 Node）；若环境已有 Node / 需要 Command Pack，再用 npm。不要臆造版本号或路径；以用户环境为准。

-## 1. 安装前检查
+## 1. 推荐：二进制安装（无需 Node）

-**原则**：先执行检查；若 **Node / 包管理器 / 网络** 任一不满足下文要求，Agent 应 **在本机终端内主动完成安装或配置**（例如升级 Node、切换镜像、配置 `npm` 代理等），再继续 `bailian-cli` 安装。仅在无权限或环境无法自动化时，再向用户说明原因并请其配合。
+```bash
+curl -fsSL https://bailian.aliyun.com/cli/install.sh | bash
+```

-1. **Node.js**
-   - 要求 **≥ 18.17.0**，执行 `node -v` 确认。
-   - 未安装或版本不足：由 Agent 自行安装或升级 Node。
+可选：正式安装读 CDN 上的 `manifest.json`。预发 / channel 验证一律读 `sync-release.json`（bailian-cli 的 channel 发版都会覆盖它）：

-2. **npm**（**唯一**允许的包管理器）
-   - 执行 `npm -v` 确认可用。
-   - 未安装或不可用：由 Agent **安装或修复 npm**（不要改用 pnpm/yarn 装 `bailian-cli`），再执行 `npm install -g bailian-cli`。
-   - 即使本机已安装 `pnpm`、`yarn` 等，**也仅用于其它项目**；装 `bailian-cli` 时仍必须用 `npm`。
+```bash
+curl -fsSL https://bailian.aliyun.com/cli/install.sh | bash -s -- --channel sync-release
+```

-3. **网络**
-   - 需能访问 npm registry（默认 **registry.npmjs.org**）。
-   - 不可达：由 Agent 自行配置镜像或代理后重试。
+也可用 `--version` / `--cdn`（或环境变量 `BAILIAN_CLI_CDN`）覆盖资源根。安装脚本本身不在本仓库维护。

---
+Windows PowerShell：

-## 2. 安装 CLI 与 Skills
+```powershell
+# 正式（manifest.json）
+irm https://bailian.aliyun.com/cli/install.ps1 | iex

-**仅允许以下命令**（不要用 `pnpm add -g`、`yarn global add` 等）。按顺序执行，上一步通过后再进行下一步。
+# channel / 预发验证（sync-release.json）
+$env:BAILIAN_CHANNEL = 'sync-release'; irm 'https://bailian.aliyun.com/cli/install.ps1' | iex
+```

-**2.1 安装 CLI**
+带其它参时先落盘再执行（`irm | iex` 不便传参），或使用仓外静态资源文档中的预发入口。
+
+二进制安装布局为 `versions/<ver>/` + `current` 指针；`bl update` 只切换指针并清理旧版本（保留当前与上一版）。更新进程退出后，下次执行 `bl` 即使用新版本（无需「重启应用」）。
+
+校验：
+
+```bash
+bl --version
+which bl   # Windows: where.exe bl
+```
+
+> CDN / GitHub Release 未就绪或下载失败时，回退到下方 npm 安装。
+
+## 2. 备选：npm 安装（要求 **≥ 18.17.0**）
+
+1. `node -v` 确认版本。
+2. `npm -v` 确认可用（**仅允许 npm** 全局安装，不要用 pnpm/yarn）。
+3. 执行：

 ```bash
 npm install -g bailian-cli
 ```

-安装成功后，应能在 PATH 中找到：
+4. 校验：`bl --version`。

- `bl`（短别名）
- `bailian`（全名）
-
-**校验**（Agent 应执行并检查退出码与输出）：
-
-```bash
-bl --version
-which bl   # Windows 可用 where bl
-```
-
-若 `command not found`：检查全局 bin 是否在 PATH（`npm config get prefix`，其下 `bin` 目录应加入 PATH）。
-
-**2.2 安装 Skills**
-
-CLI 校验通过后，在本机终端执行：
+可选 skills（与 CLI 本体无关，按需）：

 ```bash
 npx skills add modelstudioai/cli --all -g
 ```

+**Supported：** 始终使用 `--all -g`，一次装齐整套 `bailian-*`（含共享协议 `bailian-protocol`）。Agent Skills / `npx skills` **不会**按 metadata 自动拉依赖。
+
+**Advanced / 不推荐：** 子集 `-s` 时 skills CLI 不会自动带上 `bailian-protocol`；若坚持子集，必须手动同时指定，例如：
+
+```bash
+# Advanced: you MUST include bailian-protocol yourself — installer does not pull it
+npx skills add modelstudioai/cli -g -s bailian-protocol -s bailian-gen
+```
+
 安装成功后，用中文简要说明已安装的 skills 及用户可做什么。

 ---
@@ -61,39 +74,14 @@ npx skills add modelstudioai/cli --all -g

 ### 推荐：浏览器登录（控制台会话）

-适用于本机交互式安装，无需用户手动复制 API Key：
-
 1. 执行 `bl auth status --output json`，判断是否已配置。
-2. 若未配置，在**用户本机终端**执行 `bl auth login --console`；命令会拉起浏览器完成阿里云控制台登录授权。
+2. 若未配置，在**用户本机终端**执行 `bl auth login --console`。
 3. 登录成功后执行 `bl auth status --output json` 确认；汇报时只使用 masked 字段，**禁止**回显完整凭据。

-> 此方式同时打通 `app list`、`usage free` 等控制台能力，并自动配置 API Key 调用所需的鉴权信息。
+### 备选：API Key / Token Plan

-### 备选一：由 Agent 引导用户输入普通 API Key 后登录
-
-适用于无法拉起浏览器的对话式安装（远程 SSH、CI 调试、纯终端环境等）：
-
- 获取入口：[百炼控制台 API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/api-key)
-
-1. 执行 `bl auth status --output json`，判断是否已配置。
-2. 若未配置或后续 API 校验失败，**请用户粘贴 API Key**（可说明从上述控制台复制；勿要求用户发到公开渠道）。
-3. 用户提供了 Key 之后，在**用户本机终端**执行（Agent 用终端工具跑，勿把 Key 写进回复正文）：`bl auth login --api-key <用户提供的_Key>`
-4. 登录成功后执行 `bl auth status --output json` 确认；汇报时只使用 masked 字段，**禁止**回显完整 Key。
-
-### 备选二：使用 Token Plan API Key
-
- 获取入口：[Token Plan 订阅详情](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)
-
-1. 请用户从订阅详情页获取或复制 Token Plan API Key，勿要求用户发到公开渠道。
-2. 在用户本机终端执行：`bl auth login --config token-plan --api-key <用户提供的_Key>`。
-3. `token-plan` Profile 已内置默认 Base URL；登录命令会先测试 Key，通过后才保存并激活该 Profile，无需另行配置或重复测试。
-4. 执行 `bl auth status --config token-plan --output json` 确认；汇报时只使用 masked 字段。
-
-### 其他方式
-
- **环境变量**（不落盘到配置文件）：在 shell 中配置 API Key 环境变量；变量名见 `bl auth status --help`，勿在对话中向用户解释底层命名。
- **写入配置文件**（持久化，与 `auth login` 落盘相同）：`bl config set --key api_key --value <key>`（`--key api-key` 亦可）。**不会**像 `bl auth login --api-key` 那样先校验 Key 是否可用；Agent 引导安装时仍**优先**用 `auth login`。
- **命令行临时传入**：需要 API Key 的 `bl` 子命令可在**当次**执行附加全局 `--api-key <key>`，仅本次生效、不落盘（例：`bl text chat --api-key sk-xxx --message "你好"`）。与上文持久化方式不是同一用途。
+- 普通 Key：`bl auth login --api-key <Key>`
+- Token Plan：`bl auth login --config token-plan --api-key <Key>`

 ### Agent 安全约束

@@ -104,22 +92,16 @@ npx skills add modelstudioai/cli --all -g

 ## 4. 配置验证

-API Key 登录命令本身已经完成可用性测试，通过后只需确认配置状态：
-
 ```bash
 bl auth status --output json
 ```

-无需再执行重复的模型调用测试。若登录失败，根据 stderr / JSON 中的 `hint` 或 `message` 排查（网络、Key 无效、`base_url` 等）。DashScope 端点：使用 `--base-url` / `bl config set --key base_url` / `DASHSCOPE_BASE_URL`，默认中国大陆 `https://dashscope.aliyuncs.com`。
+## 5. 常见问题

---
-
-## 5. 常见问题（Agent 排障清单）
-
-| 现象                    | 可能原因             | 建议动作                                                        |
-| ----------------------- | -------------------- | --------------------------------------------------------------- |
-| `bl: command not found` | 全局 bin 不在 PATH   | 检查 `npm prefix -g` 与 PATH                                    |
-| 安装报错 engines        | Node 版本过低        | 升级到 ≥ 18.17                                                  |
-| 401 / 鉴权失败          | 未 login 或 Key 无效 | 按 Key 类型重新执行普通或 Token Plan 登录命令                   |
-| 企业网络无法访问 npm    | 代理 / 镜像          | 配置 registry 或代理后再装                                      |
-| 本机只有 pnpm、没有 npm | Agent 误用 pnpm 安装 | 先装/修好 **npm**，再用 `npm install -g bailian-cli`；勿用 pnpm |
+| 现象                     | 可能原因                     | 建议动作                                         |
+| ------------------------ | ---------------------------- | ------------------------------------------------ |
+| `bl: command not found`  | bin 不在 PATH                | 检查 `~/.local/bin` 或 `npm prefix -g`           |
+| curl 安装 404            | GitHub Release 资产未上传    | 改用 `npm install -g bailian-cli`                |
+| Windows `bl update` 失败 | 旧布局 / 文件锁 / 网络       | 重跑 `irm .../install.ps1 \| iex` 迁移布局后重试 |
+| `plugin` 需要 npm        | 二进制安装无本机 npm         | 安装 Node，或改用 npm 版 CLI                     |
+| 安装报错 engines         | Node 版本过低（仅 npm 路径） | 升级到 ≥ 18.17.0                                 |
@@ -13,8 +13,9 @@

 ---

-_Chat with Qwen, generate images & videos, understand images, call agents,_
-_manage memory, search the web — all from your terminal._
+_Chat with Qwen, generate and edit images and videos, understand images, synthesize_
+_and recognize speech, call apps, manage memory, retrieve knowledge, search the web —_
+_every AI capability, one command away._

 _Built for AI Agents. Every command works as a structured tool call._

@@ -22,28 +23,16 @@ _Built for AI Agents. Every command works as a structured tool call._

 ## Features

-Equip your AI Agent out-of-the-box with these capabilities, composable across complex tasks:
+- **Model generation** — Full-modality generation across text, image, video, and speech, with editing and reference-based generation
+- **Asset understanding** — Parse and ask questions about images, documents, audio, and long videos
+- **App orchestration** — Call Managed Agents, agents, and workflows published on Aliyun Model Studio, wired to knowledge bases, memory, web search, and MCP tools
+- **Training & deployment** — Validate and upload datasets, fine-tune models, deploy dedicated models as endpoints
+- **Account operations** — Login, UI-based configuration, model marketplace, usage and quota, rate-limit increases, team seat management
+- **Plan onboarding** — Connect subscription plans such as Token Plan to the CLI and common coding agents in one step

- **Text chat** — Qwen3.7-max: major gains in agentic coding, frontend coding, and vibe coding
- **Multimodal (Omni)** — Full omni-modal support across text + image + audio + video
- **Image generation & editing** — Qwen-Image 2.0: pro text rendering, photorealism, strong semantic adherence, multi-image composition
- **Video generation & editing** — happyhorse-1.1 series: text-/image-/reference-to-video and natural-language video editing (up to 9-image reference)
- **Speech synthesis & recognition** — CosyVoice streaming TTS, voice cloning from 5–20s samples; FunAudio-ASR covers 30 languages including 7 Chinese dialects and 20+ Mandarin accents
- **Image & video understanding** — Qwen-VL: long-form video analysis, chart/document parsing, visual reasoning, multilingual OCR
- **Coding agent setup** — Configure Claude Code, Qwen Code, OpenCode, OpenClaw, Hermes Agent, or Codex to use DashScope with `bl config agent`
+> **Note:** App orchestration, training & deployment, account operations, and plan onboarding are currently available only to China site (aliyun.com) account holders and are not yet supported for international / global site accounts.

-> **Note:** The features below are currently available only to China site (aliyun.com) account holders and are not yet supported for international / global site accounts.
-
- **Knowledge base & memory** — Multimodal RAG retrieval and cross-session memory for personalized, coherent dialogue
- **App calls** — Invoke agents and workflows already published on Aliyun Model Studio
- **MCP integration** — Orchestrate Bailian MCP servers: list services, inspect tools, and invoke any tool directly from the terminal
- **Web search** — Real-time internet retrieval for up-to-date, accurate answers
- **Model recommendation** — Describe your scenario and get best-fit model suggestions; supports scoped search, model comparison, and alternative discovery
- **Fine-tuning & deployment** — Upload datasets, create text/audio/image fine-tune jobs (`finetune text|audio|image create`; text covers SFT/LoRA/DPO/CPT), probe job status non-blockingly (`finetune watch`), query per-model training capability (`finetune capability`), and deploy trained models as endpoints (`deploy text|audio|image create`)
- **Console capabilities** — Browse the model marketplace (`model list`) and Bailian apps (`app list`), review a unified usage view (`usage summary`), check free-tier quota (`usage free`), view model usage statistics (`usage stats`), manage workspaces (`workspace list`), and manage rate limits (`quota list/request/check/history`)
- **Local file auto-upload** — Every URL parameter accepts a local path; uploaded to free temp storage with 48-hour validity
-
-## Showcase: One-Sentence Cinematic Video
+## Showcase 1: A Cinematic Short Film from One Sentence

 <p align="center">
  <a href="https://cloud.video.taobao.com/vod/dS2F4huqbw5Nfe5L3wwb3grz2q2DNYD3retq8dU-iHo.mp4">
@@ -56,26 +45,45 @@ Equip your AI Agent out-of-the-box with these capabilities, composable across co
 A complete **2-minute, 16:9 cinematic short film** — produced end-to-end from a single natural-language sentence, with **zero manual editing**. This showcase demonstrates how an AI Agent can compose a multi-step creative pipeline by orchestrating three primitives:

 - **[Qwen Code](https://github.com/QwenLM/qwen-code)** — the agentic coding model that interprets the user's intent and drives the workflow
- **[Aliyun Model Studio CLI](https://bailian.console.aliyun.com/cli?source_channel=cli_github&)** — invokes **HappyHorse 1.1**, Aliyun Model Studio's text-/image-/reference-to-video generation model
+- **[Aliyun Model Studio CLI](https://github.com/modelstudioai/cli/)** — invokes **HappyHorse 1.1**, Aliyun Model Studio's text-/image-/reference-to-video generation model
 - **[spark-video Skill](https://github.com/JohnKeating1997/spark-video)** — handles scene decomposition, storyboarding, shot continuity, and final stitching

 ### The single prompt

 > _"Generate a roughly 2-minute video in Japanese cinematic style — a sweet, innocent first-love story about a high-school girl. The plot should be heart-fluttering enough to make viewers want to fall in love. Aspect ratio: 16:9."_
->
-> _(Original: "帮我生成一段日系影视风格，高中女生的青涩初恋故事，剧情高甜，让人看了想谈恋爱，2分钟左右的视频，尺寸是16:9")_

-### How it works
+## Showcase 2: A Short-Film Director Managed Agent from One Sentence

-1. **Qwen Code** parses the request, plans the narrative beats, and decides which tools to call.
-2. The **spark-video Skill** breaks the story into shots, writes per-shot prompts, and enforces visual continuity (characters, lighting, palette, lens language).
-3. **`bl video generate`** dispatches each shot to **HappyHorse 1.1** in parallel.
-4. The skill stitches all clips back together into a single 16:9 / ~2-min deliverable.
+<p align="center">
+  <a href="https://cloud.video.taobao.com/vod/2v0GYLbJSQb2saj4iopTJDW3iRIHsintYlK-wTKbhqE.mp4">
+    <img src="https://img.alicdn.com/imgextra/i4/6000000001674/O1CN01xhzixhxltbH3LxWu_!!6000000001674-0-tbvideo.jpg" alt="Click to play the demo video" width="720" />
+  </a>
+</p>

-No timeline scrubbing. No frame-by-frame editing. Just one sentence → one video.
+<p align="center"><i>👆 Click the cover to play the full demo</i></p>
+
+One sentence builds a reusable cloud-side short-film director for storyboarding, storyboard image generation, and video creation:
+
+- **[Qwen Code](https://github.com/QwenLM/qwen-code)** — understands the requirement and generates the agent configuration
+- **[Aliyun Model Studio CLI](https://github.com/modelstudioai/cli/)** — validates the configuration, previews the changes, and completes the deployment
+- **[Managed Agent](https://bailian.console.aliyun.com/cn-beijing/?tab=managed-agents#/managed-agents/quick-start)** — runs the director role along with its skills and tools in the cloud
+
+### The single prompt
+
+> _"Build me a Managed Agent app that can produce short films — a director expert that generates videos and can also design the matching storyboards."_

 ## Installation

+**Agent install (recommended)**
+
+Send the following to your Agent — it will detect your environment, then install and verify the CLI for you:
+
+```text
+Please read https://bailian.aliyun.com/cli/install.md and install the Aliyun Model Studio CLI for me
+```
+
+**Manual install (npm)**
+
 ```bash
 npm install -g bailian-cli
 npx skills add modelstudioai/cli --all -g
@@ -85,91 +93,29 @@ npx skills add modelstudioai/cli --all -g

 ## Quick Start

-```bash
-# Authenticate, recommended
-bl auth login --console
+Once installed, just describe your task to your AI Agent — no need to assemble commands by hand.

-# Or authenticate with an API key
-bl auth login --api-key sk-xxxxx
-
-# Or use Token Plan (Base URL built in; the key is tested during login)
-bl auth login --config token-plan --api-key sk-sp-xxxxx
-
-# Configure a coding agent to use DashScope
-bl config agent --agent codex --base-url https://dashscope.aliyuncs.com/compatible-mode/v1 --api-key sk-xxxxx --model qwen3-coder-plus
-
-# Chat with Qwen
-bl text chat --message "What is DashScope?"
-
-# Multimodal chat (text + image + audio + video)
-bl omni --message "Describe this image" --image ./photo.jpg
-
-# Generate an image
-bl image generate --prompt "A cat in a spacesuit" --out-dir ./images/
-
-# Generate a video from local image
-bl video generate --image ./cat.png --prompt "Make the cat move" --download cat.mp4
-
-# Model recommendation — find the best model for your use case
-bl advisor recommend --message "I need a visual-understanding chatbot"
-
-# Compare specific models
-bl advisor recommend --message "qwen-max vs deepseek-v3 for code generation"
-
-# Browser login (required for console capability commands)
-bl auth login --console
-
-# Fine-tune & deploy — a one-shot train-to-serve workflow
-bl dataset upload --file ./train.jsonl                 # Upload a .jsonl dataset (validated first)
-bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft-lora  # Local paths auto-upload
-bl finetune watch --job-id ft-xxx --output json       # Non-blocking probe (running/succeeded return 0; failed/canceled report an error)
-bl finetune capability --model qwen3-8b               # Which training types a model supports
-bl deploy text create --model qwen3-8b --name my-svc --plan mu  # Deploy the trained model as an endpoint
-
-# Browse models / apps / free-tier quota / usage statistics / workspaces
-bl model list                                        # Browse model families and pricing
-bl app list
-bl usage summary                                     # Unified view: free-tier quota + recent usage overview
-bl usage free                                         # Free-tier quota across models (add --model/--expiring/--sort)
-bl usage stats --workspace-id <id>                    # Model usage statistics (add --model for per-model)
-bl workspace list                                     # List all workspaces
-
-# Rate limit management (list / check / request / history)
-bl quota list                                         # View RPM/TPM limits (add --model to filter)
-bl quota check                                        # Current usage vs rate limits (add --model/--period)
-bl quota request --model qwen3.6-plus --tpm 6000000   # Request a temporary TPM increase
-bl quota history                                      # View quota-change history
-
-# Token Plan team management (requires AK/SK, see auth below)
-bl token-plan list-seats                                # View subscription seat details
-bl token-plan add-member --account-name dev --org-id org_xxx
-bl token-plan assign-seats --workspace-id ws_xxx --seat-type standard --account-id acc_xxx
-bl token-plan create-key --account-id acc_xxx --workspace-id ws_xxx
-```
+| Scenario                 | What to say to your Agent                                                         |
+| ------------------------ | --------------------------------------------------------------------------------- |
+| Managed Agent            | "Create a Managed Agent that can generate short-film storyboards and videos."     |
+| Image & video generation | "Generate an image of a cat in a spacesuit on Mars, then turn it into a video."   |
+| Usage & quota            | "Show my recent model usage, free-tier quota, and rate limits."                   |
+| Model selection          | "Recommend a model for image understanding and customer support."                 |
+| About Bailian CLI        | "Tell me what Bailian CLI can do for me, and suggest how to use it for my needs." |

 > More examples and scenarios: [Aliyun Model Studio CLI Site](https://bailian.console.aliyun.com/cli?source_channel=cli_github&)

 ## Authentication

-### DashScope API Key
+### API Key

 Required for most commands. Get your key from the [DashScope Console](https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key).

 ```bash
-# Option 1: Environment variable
-export DASHSCOPE_API_KEY=sk-xxxxx
-
-# Option 2: Login command (persisted to ~/.bailian/config.json)
 bl auth login --api-key sk-xxxxx
-
-# Option 3: Per-command flag
-bl text chat --api-key sk-xxxxx --message "Hello"
 ```

-### Token Plan API Key
-
-Get or copy the API key from the [Token Plan subscription overview](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview).
-The CLI has the default Token Plan Base URL built in. Login tests the key first, then saves and activates the `token-plan` config only when validation succeeds.
+Get or copy your Token Plan API key from the [Token Plan subscription overview](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview).

 ```bash
 bl auth login --config token-plan --api-key sk-sp-xxxxx
@@ -177,26 +123,20 @@ bl auth login --config token-plan --api-key sk-sp-xxxxx

 ### Console Login (OAuth)

-Required for console capability commands (`model list`, `app list`, `usage summary/free/stats`, `workspace list`, `quota list/request/check/history`). Opens the Bailian console in your browser to sign in.
+Required for console capability commands (model list, app list, MCP list, workspace, usage queries, rate-limit increases, direct console calls). Opens the Bailian console in your browser to sign in.

 ```bash
 bl auth login --console
 ```

-### Alibaba Cloud OpenAPI AK/SK (Token Plan only)
+### Alibaba Cloud OpenAPI AK/SK

-Required for the `token-plan` command group. Get your AccessKey from [RAM Console](https://ram.console.aliyun.com/manage/ak).
+Token Plan seat and member management requires an Alibaba Cloud AccessKey. Get yours from the [RAM Console](https://ram.console.aliyun.com/manage/ak).

 > Recommended: create a RAM sub-account with minimum privileges instead of using the root account's AK/SK.

 ```bash
-# Option 1: Login command (persisted to ~/.bailian/config.json)
 bl auth login --open-api --access-key-id LTAI5t... --access-key-secret ...
-
-# Option 2: Environment variables
-export ALIBABA_CLOUD_ACCESS_KEY_ID=LTAI5t...
-export ALIBABA_CLOUD_ACCESS_KEY_SECRET=...
-export BAILIAN_WORKSPACE_ID=ws-...
 ```

 ## Configuration
@@ -205,17 +145,31 @@ export BAILIAN_WORKSPACE_ID=ws-...
 # View current config
 bl config show

-# Set defaults
-bl config set --key base_url --value https://dashscope-us.aliyuncs.com
-bl config set --key default_text_model --value qwen-turbo
-bl config set --key timeout --value 600
+# List all config profiles
+bl config list

-# Self-update to latest version
-bl update
+# Switch config profile
+bl config use --name token-plan
 ```

 Config file location: `~/.bailian/config.json`

+## Update
+
+```bash
+bl update
+```
+
+Upgrades the CLI to the latest version and refreshes the installed Agent Skills. Release notes for every version live in [CHANGELOG.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG.md).
+
+## Contributing
+
+Bug reports, feature requests, and PRs are welcome. See [CONTRIBUTING.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING.md) for developer setup, repo layout, and the workflow for adding or changing commands.
+
+Scan the QR code to join the Aliyun Model Studio CLI DingTalk user group for usage help, troubleshooting, bug reports, and tips from other users.
+
+<img src="https://img.alicdn.com/imgextra/i3/O1CN015uuhYGb6j0L12xJZ_!!6000000006304-2-tps-516-485.png" alt="Aliyun Model Studio CLI DingTalk user group" width="240" />
+
 ## Links

 | Resource                     | URL                                                                                       |
@@ -227,11 +181,3 @@ Config file location: `~/.bailian/config.json`
 | Get API Key                  | https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key |
 | Get Token Plan API Key       | https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview         |
 | Get AccessKey                | https://ram.console.aliyun.com/manage/ak                                                  |
-
-## Changelog
-
-Release notes for every version live in [CHANGELOG.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG.md).
-
-## Contributing
-
-Bug reports, feature requests, and PRs are welcome. See [CONTRIBUTING.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING.md) for developer setup, repo layout, and the workflow for adding or changing commands.
@@ -22,28 +22,16 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_

 ## 功能特性

-让您的 AI Agent 开箱即具备以下能力，并可在复杂任务中自动组合调用：
+- **模型生成** — 文本、图像、视频、语音全模态生成，支持编辑与参考生成
+- **素材理解** — 图像、文档、音频、长视频的解析与问答
+- **应用编排** — 调用百炼已发布的 Managed Agent、智能体和工作流，接入知识库、记忆库、联网搜索与 MCP 工具
+- **模型训推** — 数据集校验上传、模型精调、专属模型部署上线
+- **账号运维** — 授权登录、界面化配置、模型市场、用量与额度、限流提额、团队席位管理
+- **套餐接入** — 支持 Token Plan 等订阅计划一键接到 CLI 和常见 Coding Agent

- **文本对话** — Qwen3.7-max：Agentic coding、前端编程、Vibe coding 等能力显著增强
- **全模态对话** — 文本 + 图像 + 音频 + 视频全模态支持
- **图像生成与编辑** — Qwen-Image 2.0：专业文字渲染、真实质感、强语义遵循、多图合成
- **视频生成与编辑** — happyhorse-1.1 系列，支持文生 / 图生 / 参考生（最多 9 张图参考）/ 自然语言视频编辑
- **语音合成与识别** — CosyVoice 实时流式合成，5-20s 样本即可克隆；FunAudio-ASR 覆盖 30 种语种，含汉语七大方言与 20+ 口音官话
- **图像与视频理解** — Qwen-VL：长视频解析、复杂图表与文档识别、视觉推理、多语种 OCR
- **Coding Agent 配置** — 使用 `bl config agent` 将 Claude Code、Qwen Code、OpenCode、OpenClaw、Hermes Agent 或 Codex 配置为使用 DashScope
+> **注意：** 应用编排、模型训推、账号运维和套餐接入目前仅支持中国站（aliyun.com）账号，暂不支持国际站 / 全球站账号。

-> **注意：** 以下功能目前仅对中国站（aliyun.com）账号开放，国际站 / 全球站账号暂不支持。
-
- **知识库与记忆库** — 多模态 RAG 检索 + 跨会话记忆，提供个性化连贯对话体验
- **应用调用** — 调用已发布在阿里云百炼平台上的智能体与工作流应用
- **MCP 集成** — 统一调度百炼 MCP 服务：列出服务、查看工具、直接在终端调用任意工具
- **联网搜索** — 实时互联网信息检索，提升回答准确性及时效性
- **模型推荐** — 描述你的场景，智能推荐最适合的模型；支持限定范围搜索、模型对比和替代发现
- **微调与部署** — 上传数据集、创建文本/音频/图像调优任务（`finetune text|audio|image create`；文本涵盖 SFT/LoRA/DPO/CPT）、非阻塞探测任务状态（`finetune watch`）、按模型查训练能力（`finetune capability`），并把训练好的模型部署为推理服务（`deploy text|audio|image create`）
- **控制台能力** — 浏览模型市场（`model list`）和百炼应用（`app list`），查看统一用量视图（`usage summary`），查询模型免费额度（`usage free`），查看模型用量统计（`usage stats`），管理业务空间（`workspace list`），管理限流与提额（`quota list/request/check/history`）
- **本地文件自动上传** — 所有 URL 参数同时支持本地路径，免费临时存储 48 小时
-
-## 示例:一句话生成一部电影短片
+## 示例 1：一句话生成一部电影短片

 <p align="center">
  <a href="https://cloud.video.taobao.com/vod/dS2F4huqbw5Nfe5L3wwb3grz2q2DNYD3retq8dU-iHo.mp4">
@@ -53,27 +41,48 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_

 <p align="center"><i>👆 点击封面播放完整 2 分钟演示</i></p>

-一部完整的 **2 分钟、16:9 电影感短片** —— 由一句自然语言端到端生成,**全程零手动剪辑**。这个示例展示了 AI Agent 如何把三个基础能力编排成一条多步创作流水线:
+一部完整的 **2 分钟、16:9 电影感短片** —— 由一句自然语言端到端生成，**全程零手动剪辑**。这个示例展示了 AI Agent 如何把三个基础能力编排成一条多步创作流水线：

- **[Qwen Code](https://github.com/QwenLM/qwen-code)** —— Agentic coding 模型,解析用户意图、驱动整个工作流
- **[阿里云百炼 CLI](https://github.com/modelstudioai/cli/)** —— 调用 **HappyHorse 1.1**,百炼的文生/图生/参考生视频模型
+- **[Qwen Code](https://github.com/QwenLM/qwen-code)** —— Agentic coding 模型，解析用户意图、驱动整个工作流
+- **[阿里云百炼 CLI](https://github.com/modelstudioai/cli/)** —— 调用 **HappyHorse 1.1**，百炼的文生/图生/参考生视频模型
 - **[spark-video Skill](https://github.com/JohnKeating1997/spark-video)** —— 负责场景拆分、分镜设计、镜头连贯性和最终拼接

 ### 唯一的提示词

-> _"帮我生成一段日系影视风格,高中女生的青涩初恋故事,剧情高甜,让人看了想谈恋爱,2 分钟左右的视频,尺寸是 16:9"_
+> _“帮我生成一段日系影视风格，高中女生的青涩初恋故事，剧情高甜，让人看了想谈恋爱，2 分钟左右的视频，尺寸是 16:9。”_

-### 工作流程
+## 示例 2：一句话构建短片导演 Managed Agent

-1. **Qwen Code** 解析需求、规划叙事节奏,决定要调用哪些工具。
-2. **spark-video Skill** 把故事拆成镜头、为每个镜头写提示词,并保证视觉连贯性(角色、光线、色调、镜头语言)。
-3. **`bl video generate`** 把每个镜头并行下发给 **HappyHorse 1.1**。
-4. Skill 把所有片段拼成最终的 16:9 / 约 2 分钟成片。
+<p align="center">
+  <a href="https://cloud.video.taobao.com/vod/2v0GYLbJSQb2saj4iopTJDW3iRIHsintYlK-wTKbhqE.mp4">
+    <img src="https://img.alicdn.com/imgextra/i4/6000000001674/O1CN01xhzixhxltbH3LxWu_!!6000000001674-0-tbvideo.jpg" alt="点击播放演示视频" width="720" />
+  </a>
+</p>

-没有时间线拖拽,没有逐帧剪辑。一句话 → 一部短片。
+<p align="center"><i>👆 点击封面播放完整演示</i></p>
+
+一句话构建一个可复用的云端短片导演，用于分镜设计、分镜图生成和视频创作：
+
+- **[Qwen Code](https://github.com/QwenLM/qwen-code)** —— 理解需求并生成 Agent 配置
+- **[阿里云百炼 CLI](https://github.com/modelstudioai/cli/)** —— 校验配置、预览变更并完成部署
+- **[Managed Agent](https://bailian.console.aliyun.com/cn-beijing/?tab=managed-agents#/managed-agents/quick-start)** —— 在云端运行导演角色及其 Skill 和工具
+
+### 唯一的提示词
+
+> _“帮我构建一个 managedagent 应用，能够实现短片拍摄，导演专家生成视频，然后也能进行设计对应的分镜图。”_

 ## 安装

+**Agent 安装（推荐）**
+
+把下面这句话发给你的 Agent，它会自行判断环境并完成安装与校验：
+
+```text
+请阅读：https://bailian.aliyun.com/cli/install.md 并按照说明为我安装阿里云百炼 CLI
+```
+
+**手动安装（npm）**
+
 ```bash
 npm install -g bailian-cli
 npx skills add modelstudioai/cli --all -g
@@ -83,91 +92,29 @@ npx skills add modelstudioai/cli --all -g

 ## 快速开始

-```bash
-# 认证（推荐浏览器登录）
-bl auth login --console
+安装完成后，直接在 AI Agent 中描述你的任务，无需手动拼接命令。

-# 或使用 API key 认证
-bl auth login --api-key sk-xxxxx
-
-# 或使用 Token Plan（已内置 Base URL，登录时自动测试 Key）
-bl auth login --config token-plan --api-key sk-sp-xxxxx
-
-# 配置 Coding Agent 使用 DashScope
-bl config agent --agent codex --base-url https://dashscope.aliyuncs.com/compatible-mode/v1 --api-key sk-xxxxx --model qwen3-coder-plus
-
-# 和通义千问对话
-bl text chat --message "你好，介绍一下阿里云百炼平台"
-
-# 多模态对话（文本 + 图片 + 音频 + 视频）
-bl omni --message "描述这张图片" --image ./photo.jpg
-
-# 生成图片
-bl image generate --prompt "一只穿太空服的猫在火星上" --out-dir ./images/
-
-# 图生视频（本地文件自动上传）
-bl video generate --image ./cat.png --prompt "让画面中的猫动起来" --download cat.mp4
-
-# 模型推荐 — 根据场景推荐最适合的模型
-bl advisor recommend --message "我要做一个能理解图片的客服机器人"
-
-# 对比特定模型
-bl advisor recommend --message "qwen-max 和 deepseek-v3 哪个更适合做代码生成"
-
-# 浏览器登录（控制台能力相关命令需要）
-bl auth login --console
-
-# 微调与部署 — 从训练到服务的一站式流程
-bl dataset upload --file ./train.jsonl                 # 上传 .jsonl 数据集（先校验）
-bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft-lora  # 本地路径自动上传
-bl finetune watch --job-id ft-xxx --output json       # 非阻塞探测（运行中/成功返回 0；失败/取消报错）
-bl finetune capability --model qwen3-8b               # 查询模型支持哪些训练方式
-bl deploy text create --model qwen3-8b --name my-svc --plan mu  # 把训练好的模型部署为推理服务
-
-# 浏览模型 / 应用 / 免费额度 / 用量统计 / 业务空间
-bl model list                                        # 浏览模型系列与价格信息
-bl app list
-bl usage summary                                     # 统一视图：免费额度 + 近期用量概览
-bl usage free                                         # 各模型免费额度（可加 --model/--expiring/--sort）
-bl usage stats --workspace-id <id>                    # 模型用量统计（加 --model 查单模型）
-bl workspace list                                     # 列出所有业务空间
-
-# 限流管理与提额（list / check / request / history）
-bl quota list                                         # 查看 RPM/TPM 限额（加 --model 过滤）
-bl quota check                                        # 当前用量 vs 限流阈值（加 --model/--period）
-bl quota request --model qwen3.6-plus --tpm 6000000   # 申请临时 TPM 提额
-bl quota history                                      # 查看提额历史记录
-
-# Token Plan 团队版管理（需 AK/SK，见下方认证说明）
-bl token-plan list-seats                                # 查看订阅席位明细
-bl token-plan add-member --account-name dev --org-id org_xxx
-bl token-plan assign-seats --workspace-id ws_xxx --seat-type standard --account-id acc_xxx
-bl token-plan create-key --account-id acc_xxx --workspace-id ws_xxx
-```
+| 场景             | 可以这样对 Agent 说                                                     |
+| ---------------- | ----------------------------------------------------------------------- |
+| Managed Agent    | “帮我创建一个能够生成短片分镜和视频的 Managed Agent。”                  |
+| 图片和视频生成   | “生成一张穿着太空服的猫站在火星上的图片，再把它制作成一段视频。”        |
+| 用量与额度       | “查看最近的模型用量、免费额度和限流情况。”                              |
+| 模型选型         | “推荐一个适合图片理解和智能客服的模型。”                                |
+| 了解 Bailian CLI | “介绍一下 Bailian CLI 能帮我完成哪些任务，并根据我的需求推荐使用方式。” |

 > 更多案例与使用场景：[阿里云百炼 CLI 官方主页](https://bailian.console.aliyun.com/cli?source_channel=cli_github&)

 ## 认证方式

-### DashScope API Key
+### API Key

 大部分命令均需要 API Key。前往 [DashScope 控制台](https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key) 获取。

 ```bash
-# 方式一：环境变量
-export DASHSCOPE_API_KEY=sk-xxxxx
-
-# 方式二：登录命令（持久化到 ~/.bailian/config.json）
 bl auth login --api-key sk-xxxxx
-
-# 方式三：命令行参数
-bl text chat --api-key sk-xxxxx --message "你好"
 ```

-### Token Plan API Key
-
-前往 [Token Plan 订阅详情](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) 获取或复制 API Key。
-CLI 已内置 Token Plan 的默认 Base URL；登录命令会先测试 Key，通过后才保存并激活 `token-plan` 配置。
+Token Plan 的 API Key 前往 [Token Plan 订阅详情](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) 获取或复制。

 ```bash
 bl auth login --config token-plan --api-key sk-sp-xxxxx
@@ -175,26 +122,20 @@ bl auth login --config token-plan --api-key sk-sp-xxxxx

 ### 控制台登录（OAuth）

-控制台能力命令（`model list`、`app list`、`usage summary/free/stats`、`workspace list`、`quota list/request/check/history`）需要使用此登录方式。打开浏览器跳转百炼控制台完成登录。
+控制台能力命令（模型列表、应用列表、MCP 列表、工作空间、用量查询、限流提额、控制台直调）需要使用此登录方式。打开浏览器跳转百炼控制台完成登录。

 ```bash
 bl auth login --console
 ```

-### 阿里云 OpenAPI AK/SK（仅 Token Plan）
+### 阿里云 OpenAPI AK/SK

-`token-plan` 命令组需要阿里云 AccessKey。前往 [RAM 控制台](https://ram.console.aliyun.com/manage/ak) 获取。
+Token Plan 的席位与成员管理需要阿里云 AccessKey。前往 [RAM 控制台](https://ram.console.aliyun.com/manage/ak) 获取。

 > 建议：创建 RAM 子账号并授予最小权限，避免使用主账号 AK/SK。

 ```bash
-# 方式一：登录命令（持久化到 ~/.bailian/config.json）
 bl auth login --open-api --access-key-id LTAI5t... --access-key-secret ...
-
-# 方式二：环境变量
-export ALIBABA_CLOUD_ACCESS_KEY_ID=LTAI5t...
-export ALIBABA_CLOUD_ACCESS_KEY_SECRET=...
-export BAILIAN_WORKSPACE_ID=ws-...
 ```

 ## 配置
@@ -203,17 +144,31 @@ export BAILIAN_WORKSPACE_ID=ws-...
 # 查看当前配置
 bl config show

-# 设置默认值
-bl config set --key base_url --value https://dashscope-us.aliyuncs.com
-bl config set --key default_text_model --value qwen-turbo
-bl config set --key timeout --value 600
+# 查看全部配置档
+bl config list

-# 自更新到最新版本
-bl update
+# 切换配置档
+bl config use --name token-plan
 ```

 配置文件位置：`~/.bailian/config.json`

+## 更新
+
+```bash
+bl update
+```
+
+升级 CLI 至最新版本，并同步更新已安装的 Agent Skills。每个版本的变更详情记录在 [CHANGELOG.zh.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG.zh.md)。
+
+## 参与贡献
+
+欢迎提 Issue、Feature Request 和 PR。开发环境搭建、仓库结构、新增/修改命令的工作流请见 [CONTRIBUTING.zh.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING.zh.md)。
+
+欢迎扫码加入阿里云百炼 CLI 钉钉用户交流群，获取使用答疑、问题排查、Bug 反馈和使用经验交流支持。
+
+<img src="https://img.alicdn.com/imgextra/i3/O1CN015uuhYGb6j0L12xJZ_!!6000000006304-2-tps-516-485.png" alt="阿里云百炼 CLI 钉钉用户交流群" width="240" />
+
 ## 相关链接

 | 资源                    | 地址                                                                                      |
@@ -225,11 +180,3 @@ bl update
 | 获取 API Key            | https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key |
 | 获取 Token Plan API Key | https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview         |
 | 获取 AccessKey          | https://ram.console.aliyun.com/manage/ak                                                  |
-
-## 更新日志
-
-每个版本的变更详情记录在 [CHANGELOG.zh.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG.zh.md)。
-
-## 参与贡献
-
-欢迎提 Issue、Feature Request 和 PR。开发环境搭建、仓库结构、新增/修改命令的工作流请见 [CONTRIBUTING.zh.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING.zh.md)。
@@ -25,7 +25,7 @@ defineCommand({ auth }) → runtime/authStage → ctx.client → command.run(ctx
 当前 command 鉴权域(`AuthRequirement`):

 - `apiKey` — DashScope / OpenAI-compatible 模型域,用 API key 与 model base URL
- `console` — Bailian Console Gateway,用 console access token + region/site/switchAgent/workspace
+- `console` — Bailian Console Gateway,用 console access token + region/site/switchAgent;`workspace_id` 是独立的 Settings 作用域,不属于 credential
 - `openapi` — 阿里云 OpenAPI 签名域,用 AccessKey ID/Secret 调用 Token Plan 等 OpenAPI
 - `none` — 本地命令、登录/配置类命令、无需 credential 的命令

@@ -35,7 +35,7 @@ defineCommand({ auth }) → runtime/authStage → ctx.client → command.run(ctx

 - `bl auth login --api-key ...` 只更新 `api_key` / `base_url`
 - `bl auth login --console` 只更新 `access_token` 以及回调携带的 console 作用域字段
- `bl auth login --open-api ...` 只更新 `access_key_id` / `access_key_secret`
+- `bl auth login --open-api ...` 更新 `access_key_id` / `access_key_secret`,同时会调用 OpenAPI 生成 CLI `access_token` 并一并写入;即一次 `--open-api` 登录同时产生 `openapi` 与 `console` 域凭证
 - `bl auth logout --console` 只清 `access_token`
 - `bl auth logout --open-api` 只清 `access_key_id` / `access_key_secret` / `security_token`
 - `bl auth logout` 清 `api_key` + `base_url` + `access_token` + `access_key_*`
@@ -78,6 +78,9 @@ defineCommand({ auth }) → runtime/authStage → ctx.client → command.run(ctx
  - 如新增鉴权域,扩展 `AuthRequirement`
  - 更新 `credentialFlagDefs()` 暴露该域可见的 flag
  - 必要时新增 `*_AUTH_FLAGS`
+  - `workspace_id` 是作用域字段而非 credential,不要把它放进 `ConsoleCredential`;读取方式按命令 `auth` 域区分:
+    - `auth: "console"` 命令通过 `CONSOLE_AUTH_FLAGS` 自动获得 `--workspace-id`,由 `buildSettings()` 解析到 `settings.workspaceId`,命令统一从 `settings.workspaceId` 读取
+    - `auth: "apiKey"`/`"openapi"`/`"none"` 命令如需 `--workspace-id`,必须自声明 flag;因它不会进入 credential/global flags,命令从 `ctx.flags.workspaceId` 读取(可回退到 `settings.workspaceId`)
 - [ ] `packages/core/src/auth/types.ts`:
  - 新增 credential 类型 / source / scope 字段
 - [ ] `packages/core/src/auth/resolver.ts`:
@@ -121,7 +124,7 @@ defineCommand({ auth }) → runtime/authStage → ctx.client → command.run(ctx
 ### D. 用户面文档

 - [ ] `README.md` / `README.zh.md` "Authentication" 段落
- [ ] `skills/bailian-cli/reference/` 通过 `pnpm run sync:skill-assets` 重建
+- [ ] 各 `skills/<skill>/reference/` 通过 `pnpm run sync:skill-assets` 重建

 ### E. 测试

@@ -131,6 +134,8 @@ defineCommand({ auth }) → runtime/authStage → ctx.client → command.run(ctx

 ## 完成后自查

+本仓库同时存在 `bl`(packages/cli) 与 `kscli`(packages/kscli) 两个入口,二者共享 core/runtime 鉴权链路,但暴露的命令不同。如果改动会影响两个入口共用的命令或错误提示,再分别验证它们各自实际暴露的路径;不要假设 `kscli` 也有 `bl auth *` 命令。
+
 ```sh
 # 各种凭证组合
 unset DASHSCOPE_API_KEY ALIBABA_CLOUD_ACCESS_KEY_ID ALIBABA_CLOUD_ACCESS_KEY_SECRET
@@ -150,9 +155,11 @@ Console 登录/网关相关改动:

 ```sh
 pnpm -F bailian-cli exec tsx src/main.ts auth login --console
-pnpm -F bailian-cli exec tsx src/main.ts usage stats --dry-run --output json
+pnpm -F bailian-cli exec tsx src/main.ts usage stats --dry-run --output json --workspace-id ws-xxx
 ```

+注意:`usage stats --dry-run` 仍会先校验 workspace,必须传入 `--workspace-id`(或 `BAILIAN_WORKSPACE_ID` / config `workspace_id`)。
+
 ## 常见漏点

 - ✗ 加了新 token 来源但忘了改 resolver 优先级,实际不生效
@@ -56,7 +56,7 @@ git diff --name-only <base>...<head>

 - [ ] **新命令 / 新 flag** 已同步到用户面文档:
  - [README.md](README.md) + [README.zh.md](README.zh.md)(中英文都要,常漏 `_CN`)
-  - `skills/bailian-cli/reference/` + `skills/bailian-cli/SKILL.md` 通过 `pnpm run sync:skill-assets` 更新并提交
+  - 各 `skills/<skill>/reference/` + 对应 `SKILL.md` 通过 `pnpm run sync:skill-assets` 更新并提交
 - [ ] **`bl <cmd> --help`** 文案完整:`description` / `examples` 都填了
 - [ ] **demo / quickstart**:用户可调用的新命令至少有一个示例
 - [ ] **行为变化的老命令**:在 commit message / CHANGELOG 注明用户感知的差异
@@ -95,7 +95,7 @@ describe.skipIf(<ready>)("e2e: <topic>（DashScope …）", () => {

 - [ ] `packages/commands/src/index.ts` 导出 + `packages/cli/src/commands.ts` 暴露路径 + `topic-routes.ts` 补最小路由
 - [ ] `packages/commands/tests/e2e/<topic>.e2e.test.ts`（新建或扩展）
- [ ] 若改了 `usageArgs` / `flags` / `exampleArgs`,跑 `pnpm --filter bailian-cli run generate:reference` 更新 `skills/bailian-cli/reference/` 并提交
+- [ ] 若改了 `usageArgs` / `flags` / `exampleArgs`,跑 `pnpm --filter bailian-cli run generate:reference` 更新各 `skills/<skill>/reference/` 并提交
 - [ ] 子命令 `--help`（分组 help 由 bl `registry.smoke` 覆盖）
 - [ ] skip 块：每个 required flag 缺参；可 dry-run 则加一条
 - [ ] 至少一条真实集成（或说明为何仅 smoke）；不破坏已有集成用例顺序
@@ -56,7 +56,7 @@ packages/commands/src/index.ts
 - **`packages/cli/src/commands.ts`**:`bl` 产品命令 map;新增/删除/重命名 `bl` 命令必须改这里
 - **`packages/kscli/src/main.ts`**:`kscli` 产品命令 map;只有该入口需要暴露/变更时才改
 - **`packages/runtime/src/registry.ts`**:通用 registry,从传入 map 建树;不要在这里登记业务命令
- **`tools/generate-reference.ts`**:pre-commit / `pnpm run sync:skill-assets` 时读 `packages/cli/src/commands.ts`,写 `skills/bailian-cli/reference/index.md` + `<一级命令>.md`。该目录**纳入 git**,勿手改
+- **`tools/generate-reference.ts`**:pre-commit / `pnpm run sync:skill-assets` 时读 `packages/cli/src/commands.ts`,按 `GROUP_OWNER_SKILL` 归属表分流写到各 `skills/<skill>/reference/index.md` + `<一级命令>.md`。未显式归属的一级组默认进 `bailian-cli`。各目录**纳入 git**,勿手改。新增一级命令组若应归领域 skill,记得改归属表。

 已删除/勿再引用:旧的 `packages/cli/src/commands/catalog.ts`、旧的 `packages/cli/src/commands/index.ts` catalog re-export、`packages/cli/src/registry.ts`、`skipDefaultApiKeySetup`、`ensureApiKey` 启动拦截、`config/export-schema.ts`。

@@ -87,9 +87,10 @@ packages/commands/src/index.ts

 ### C. 文档层

- [ ] 运行 `pnpm run sync:skill-assets`(或正常 `git commit` 走 pre-commit),刷新 `skills/bailian-cli/reference/` 与 `SKILL.md` 的 `metadata.version` 并提交
+- [ ] 运行 `pnpm run sync:skill-assets`(或正常 `git commit` 走 pre-commit),刷新各 `skills/<skill>/reference/` 与 `SKILL.md` 的 `metadata.version` 并提交
 - [ ] `README.md` / `README.zh.md`:Quick Start、命令一览、认证说明(用户向,与 help 对齐)
- [ ] `skills/bailian-cli/SKILL.md`:若安装说明或能力边界有变,同步更新
+- [ ] 相关 `skills/<skill>/SKILL.md`:若安装说明或能力边界有变,同步更新;新一级命令组若属领域 skill,同步改 `tools/generate-reference.ts` 的 `GROUP_OWNER_SKILL`
+- [ ] **拥有方** skill 的「When to use which command」(或等价路由表)补上新意图;hub `bailian-cli` 仅加/改 hand-off 行,**不要**把领域子命令与默认模型抄进 hub 表(约定见 [skill-change.md](skill-change.md))

 ### D. 测试层

@@ -105,7 +106,7 @@ packages/commands/src/index.ts
  - `packages/cli/src/commands.ts` map key
  - `packages/kscli/src/commands.ts` map key(如适用)
  - 用户可见 hint / README / tests
-  - `skills/bailian-cli/reference/`(重建后检查并提交)
+  - `skills/*/reference/`(重建后检查并提交)
 - [ ] 检查 `usageArgs` / `exampleArgs` 没有硬编码旧的 `bl <path>` 前缀

 ## 完成后自查
@@ -127,7 +128,9 @@ pnpm -F knowledge-studio-cli exec tsx src/main.ts <command> --help

 - ✗ 只新增 `packages/commands/src/commands/...` 文件,忘了在 `packages/commands/src/index.ts` 导出
 - ✗ 只导出了命令实现,忘了在 `packages/cli/src/commands.ts` 暴露路径 → `bl --help` 看不到
- ✗ 手改 `skills/bailian-cli/reference/*.md` → 下次 generate 被覆盖;应改 command metadata 后重新 generate 并提交
+- ✗ 手改 `skills/*/reference/*.md` → 下次 generate 被覆盖;应改 command metadata 后重新 generate 并提交
+- ✗ 新一级命令组忘改 `tools/generate-reference.ts` 的 `GROUP_OWNER_SKILL` → reference 会落到 hub `bailian-cli`(未必是预期)
+- ✗ 只改 reference / hub,忘改拥有方 skill 路由表;或把领域命令明细重新抄回 `bailian-cli` SKILL → 与 [skill-change.md](skill-change.md) 分层冲突
 - ✗ 在 `usageArgs` / `exampleArgs` 写死 `bl text chat` → `kscli` 等入口复用时 help 错
 - ✗ Console Gateway 命令忘设 `auth: "console"` → console flags / credential 注入都不生效
 - ✗ 单 action 的子组是反模式,新增时优先拍平为两级
@@ -30,7 +30,7 @@
 ### C. 文档层

 - [ ] `README.md` / `README.zh.md` 如果在示例里展示了相关命令,补充新 flag
- [ ] 跑 `pnpm --filter bailian-cli run generate:reference`,让 `skills/bailian-cli/reference/` 与命令一致(勿手改;改完提交)
+- [ ] 跑 `pnpm --filter bailian-cli run generate:reference`,让各 `skills/<skill>/reference/` 与命令一致(勿手改;改完提交)

 ### D. 测试层

@@ -43,7 +43,7 @@
 - [ ] `packages/cli/tests/e2e/command-packs.e2e.test.ts` 覆盖 help、link、执行、output/errors、凭据授权、list、remove。
 - [ ] `packages/kscli/tests/e2e/command-packs.e2e.test.ts` 覆盖统一 host 和 runtime 默认空 policy 下不暴露管理命令。
 - [ ] fixture 的包名必须在测试白名单内，且构建入口不依赖工作区运行时解析。
- [ ] 更新生成的 `skills/bailian-cli/reference/plugin.md`；公开 `README.md` / `README.zh.md` 等正式对外发布时再补。
+- [ ] 更新生成的 `skills/bailian-cli/reference/plugin.md`（或归属表指定的 skill reference）；公开 `README.md` / `README.zh.md` 等正式对外发布时再补。

 验证：

@@ -26,7 +26,8 @@

 ### C. 命令手册

- [ ] 若 `--model` 的 description 含 default,改命令后跑 `pnpm --filter bailian-cli run generate:reference` 更新 `skills/bailian-cli/reference/<group>.md` 并提交
+- [ ] 若 `--model` 的 description 含 default,改命令后跑 `pnpm --filter bailian-cli run generate:reference` 更新对应 `skills/<skill>/reference/<group>.md` 并提交
+- [ ] 同步**拥有该命令的领域 skill**「When to use which command」表中的 Default model(现主要是 `bailian-gen`;精调相关看 `bailian-finetune` 正文示例)。hub `bailian-cli` 已瘦身,一般**不必**再写领域默认模型(见 [skill-change.md](skill-change.md))

 ### D. 用户面文档

@@ -49,6 +50,7 @@ pnpm -F bailian-cli exec tsx src/main.ts <command> --model <new-model> --message

 ## 常见漏点

- ✗ 改了命令默认模型,但 SKILL.md frontmatter 仍写老型号 → AI agent 调用时仍按老型号宣传
+- ✗ 改了命令默认模型,但 SKILL.md frontmatter 或领域路由表 Default model 仍写老型号 → AI agent 调用时仍按老型号宣传
+- ✗ 只改了 `reference/` / flag description,忘改 `bailian-gen`(等) SKILL 路由表
 - ✗ 废弃模型时只删了代码,e2e 测试还在跑,CI 红
 - ✗ 新模型 endpoint 不一致,但只改了 default,没加 endpoint 分支判断
@@ -1,27 +1,53 @@
-# 发布（npm publish）
+# 发布（npm + GitHub Release 二进制）

 ## 触发条件

- 准备发布 channel（beta/mcp/plugin 等）或正式版到 npm
- 准备打 git tag
+- 准备发布 channel（mcp/plugin 等）或正式版到 npm **与** GitHub Releases 二进制
+- 准备打 git tag（仅 stable）

-## 发布方式：GitHub Actions + npm OIDC
+## 发布方式：GitHub Actions 总入口

 发版**必须**通过 CI 完成，不要本地手动 `pnpm publish`。

 入口：GitHub Actions → **Publish** workflow（`.github/workflows/publish.yml`）→ Run workflow。

+**编排关系（重要）：**
+
+```text
+publish-stable.mjs / publish-channel.mjs   ← 唯一发版入口
+        ├─ npm（pnpm publish）
+        └─ binary（lib/binary-release
+              → binary-build
+              → gh-release
+              → oss-direct-upload）
+```
+
+`tools/release/lib/binary-release.mjs` 等是实现，一般不要单独当发版入口（调试可用）。
+
 两种模式：

-| 模式    | 用途                           | 触发方式                                           |
-| ------- | ------------------------------ | -------------------------------------------------- |
-| channel | 发 channel 版本到指定 dist-tag | 选 mode=channel，填 dist-tag 名称（如 mcp/plugin） |
-| stable  | 正式发版到 latest              | 选 mode=stable，需 production environment 审批     |
+| 模式    | 用途                                                                                    | 触发方式                                     |
+| ------- | --------------------------------------------------------------------------------------- | -------------------------------------------- |
+| channel | npm dist-tag +（仅 bailian-cli）二进制 + CDN **一律**覆盖 `sync-release.json`           | mode=channel，channel 填 **npm dist-tag** 名 |
+| stable  | npm latest + GitHub Release `v<ver>` + CDN **`manifest.json`**（及 `latest.json` 别名） | mode=stable，需 production environment 审批  |
+
+可选 flag：`--skip-binary`（仅发 npm，紧急逃生）。
+
+### CDN 滚动指针（bailian-cli）
+
+| 发布模式 | CDN 指针                           | 本机安装 / 更新                                                   |
+| -------- | ---------------------------------- | ----------------------------------------------------------------- |
+| channel  | 始终覆盖 `sync-release.json`       | `BAILIAN_CHANNEL=sync-release` / `install --channel sync-release` |
+| stable   | `manifest.json`（+ `latest.json`） | 默认安装 / `bl update`（无 channel）                              |
+
+workflow 的 `channel` 输入**只决定 npm dist-tag**（如 `mcp` / `plugin` / `sync-release`），**不再**生成 `release-test.json` 这类旁路文件。

 ### channel 发布

-1. 在 GitHub 触发 Publish workflow，package 选 `bailian-cli` 或 `knowledge-studio-cli`，mode 选 `channel`，channel 填 dist-tag 名（如 `mcp`）
-2. CI 自动：生成 `0.0.0-beta-<sha7>-<date>` 版本号 → 临时 bump 对应包集合 → 自检 → 构建 → 发布到指定 dist-tag
+1. 在 GitHub 触发 Publish workflow，mode 选 `channel`，channel 填 npm dist-tag 名：
+   - **`bailian-cli`**：npm 发到该 tag；二进制同时刷新 CDN `sync-release.json`（与 tag 名无关）。本机验证：`BAILIAN_CHANNEL=sync-release`
+   - **`knowledge-studio-cli`**：仅 npm（自动跳过 binary，不碰 `sync-release.json`）
+2. CI 自动：生成 `0.0.0-beta-<sha7>-<YYYYMMDDHHMM>`（UTC 到分钟；同 commit 同分钟重跑会覆盖同号）→ 临时 bump → 自检 → **npm 发到 dist-tag** →（bailian-cli）**Bun 编二进制 + GH prerelease + 覆盖 `sync-release.json`** → 还原 package.json
 3. 对应脚本：`tools/release/publish-channel.mjs`

 ### stable 发布
@@ -29,7 +55,7 @@
 1. 确保当前 release tooling 覆盖的包(`tools/release/lib/packages.mjs`)已升到目标版本且一致;当前基础集合为 `packages/core` / `packages/runtime` / `packages/commands` / `packages/cli`，`knowledge-studio-cli` 发布会额外包含 `packages/kscli`
 2. 在 GitHub 触发 Publish workflow，package 选目标包集合，mode 选 `stable`
 3. 需要 production environment 审批人批准
-4. CI 自动：自检 → 构建 → 检查 npm 已发布版本 → 发布到 latest → 打 git tag
+4. CI 自动：自检 → **npm 发到 latest** → **推送 git tag `v<ver>`** → **Bun 编二进制并创建/更新 GitHub Release** →（bailian-cli）维护 CDN **`manifest.json`** → 完成
 5. 如果所选发布集合的当前版本已全部存在于 npm，stable 发布会失败并提示先升级版本号；如果只有部分包已发布，CI 会继续补发缺失包
 6. 对应脚本：`tools/release/publish-stable.mjs`

@@ -37,17 +63,17 @@

 两种模式都会先跑 `check.mjs`，覆盖以下检查：

-| 检查项                           | 说明                                                                                             |
-| -------------------------------- | ------------------------------------------------------------------------------------------------ |
-| `pnpm install --frozen-lockfile` | lockfile 一致性                                                                                  |
-| README 同步                      | `packages/cli/README.md` 与根 README 一致                                                        |
-| 版本号一致                       | `tools/release/lib/packages.mjs` 中待发布包集合 version 相同                                     |
-| `workspace:*` 替换               | 发布包间 workspace 依赖解析为真实版本号                                                          |
-| 构建                             | 基础发布构建 core/runtime/commands 依赖和 cli;`--knowledge` 额外构建 `knowledge-studio-cli`      |
-| 生成资产                         | 重建 `skills/bailian-cli/reference/`;非 channel 模式还同步 `skills/bailian-cli/SKILL.md` version |
-| pnpm pack                        | 打 tarball                                                                                       |
-| publint                          | 包元数据校验                                                                                     |
-| gitleaks                         | 敏感信息扫描                                                                                     |
+| 检查项                           | 说明                                                                                                            |
+| -------------------------------- | --------------------------------------------------------------------------------------------------------------- |
+| `pnpm install --frozen-lockfile` | lockfile 一致性                                                                                                 |
+| README 同步                      | `packages/cli/README.md` 与根 README 一致                                                                       |
+| 版本号一致                       | `tools/release/lib/packages.mjs` 中待发布包集合 version 相同                                                    |
+| `workspace:*` 替换               | 发布包间 workspace 依赖解析为真实版本号                                                                         |
+| 构建                             | 基础发布构建 core/runtime/commands 依赖和 cli;`--knowledge` 额外构建 `knowledge-studio-cli`                     |
+| 生成资产                         | 重建各 `skills/<skill>/reference/`;非 channel 模式还同步各 `skills/*/SKILL.md` version（含 `bailian-protocol`） |
+| pnpm pack                        | 打 tarball                                                                                                      |
+| publint                          | 包元数据校验                                                                                                    |
+| gitleaks                         | 敏感信息扫描                                                                                                    |

 本地可以 dry-run 验证：

@@ -59,7 +85,9 @@ node tools/release/publish-channel.mjs --channel test --knowledge --dry-run
 ## CI 基础设施

 - **认证**：npm OIDC Trusted Publishing（无 token），需要 `id-token: write` 权限
+- **GitHub Release**：`contents: write` + `GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}`（stable / channel 均需）
 - **Node 版本**：24（npm 11.5+ 才支持 OIDC token 交换）
+- **Bun**：`oven-sh/setup-bun`，版本钉死在 workflow 中
 - **Actions 版本**：checkout/setup-node/pnpm-action 均为 v6（Node 24 兼容）
 - **npm 配置**：当前 release tooling 发布的包(`bailian-cli-core` / `bailian-cli-runtime` / `bailian-cli-commands` / `bailian-cli` / `knowledge-studio-cli`)的 Trusted Publisher 指向 `modelstudioai/cli` 的 `publish.yml`;新增发布包时同步 npm Trusted Publisher

@@ -105,3 +133,5 @@ node tools/release/publish-channel.mjs --channel test --knowledge --dry-run
 | npm Trusted Publisher 的 workflow filename 改了没同步               | OIDC 匹配不上，publish 报 404                                                      |
 | CI 用 Node 22（npm 10）跑 publish                                   | npm 10 不支持 OIDC token 交换，publish 报 404                                      |
 | stable 发布前没有升级版本号                                         | 所选发布集合的版本已全部存在于 npm，CI 明确报错并要求先升级版本号                  |
+| channel job 缺少 `contents: write`                                  | `gh release create` 失败                                                           |
+| stable 未先推 tag 就建 Release                                      | `--verify-tag` 失败                                                                |
@@ -0,0 +1,76 @@
+# Skill 文案 / 路由 / 安装约定
+
+## 触发条件
+
+- 改 `skills/*/SKILL.md` 的 description、路由表、consent、安全闸、hand-off、references 落款
+- 调整 `bailian-protocol` 与业务 skill 的关系，或业务 skill 之间的软 hand-off 约定
+- 新增 / 拆分 / 合并 `bailian-*` 业务 skill，或改 `tools/generate-reference.ts` 的 `GROUP_OWNER_SKILL` 归属（与命令增删改交叉时两边都看）
+- 给业务 skill 补安装说明、README，或统一「勿猜 flag → `reference/`」类约定
+
+纯改生成物 `skills/*/reference/*.md`（由命令 metadata 驱动）→ 走 [command-add-remove.md](command-add-remove.md) / [command-flag-change.md](command-flag-change.md)，**不要手改 reference**。
+
+## 统一口径（安装）
+
+1. **Supported install：** `npx skills add modelstudioai/cli --all -g`（整包装齐，含 `bailian-protocol`）
+2. **`bailian-protocol` 是共享协议 skill**，业务 skill 执行前应 Read 它；Agent Skills / `npx skills` **不会**按 frontmatter 自动拉依赖
+3. **不要**在 frontmatter 写 `companions`，也不要对外说「companions = 安装器硬依赖」
+4. 子集安装（`-s`）为 **advanced / 不推荐**：skills CLI 不会自动带上 protocol；漏装会导致相对路径 Read 失败
+
+## 概念图
+
+```text
+bailian-protocol          ← 共享协议（consent / 鉴权 / 版本 / 错误上报）
+        ▲                   靠 --all -g 与业务 skill 同装；非安装器强制 companions
+        │
+┌───────┴────────┬────────────────┬──────────────────┐
+bailian-gen      bailian-finetune  bailian-managed-agent
+（领域路由表）    （领域工作流）     （IaC 安全闸）
+        │                │                  │
+        └────────────────┼──────────────────┘
+                         ▼ 软 hand-off（按 skill 名）
+                   bailian-cli（hub）
+                   hub 路由表：本职命令 + 领域 hand-off 行
+                   细节 → 各 skill reference/（生成）
+```
+
+## 必查清单
+
+### A. 分层边界
+
+- [ ] **整包装齐**：安装/升级文案主推 `--all -g`；业务 skill **不**声明 `companions`
+- [ ] **协议读取**：CRITICAL / references 可链 `../bailian-protocol/…`；若读不到 → 停止执行 `bl`，提示 `npx skills add modelstudioai/cli --all -g`
+- [ ] **软 hand-off**：兄弟业务 skill **只写 skill 名**；已安装则 Read，未安装则 `bl … --help` 或提示整包安装；**不要**把 `../bailian-gen/…` 等写成执行前提
+- [ ] **Hub vs 领域**：`bailian-cli` 的「When to use which command」只列 hub 拥有的意图；媒体 / 精调 / managed-agent 各留 hand-off 行，**不抄**领域默认模型与子命令明细
+- [ ] **渐进披露**：SKILL 写意图路由与领域硬规则；flags / usage / examples 以 `reference/` 或 `bl <command> --help` 为准，表后保留「勿猜 flag」指向句
+
+### B. 文案与落款一致性
+
+- [ ] 领域 skill（gen / finetune / managed-agent）路由或命令表后有指向 `reference/` 的句；文末 `## references`（protocol + reference）与家族对齐
+- [ ] description 含 WHAT + WHEN + 反触发；安装说明指向 `--all -g`，不写 companions 必装
+- [ ] Quick examples 只演示本 skill 职责（hub 不示范 `bl image` / `bl video` 等）
+- [ ] 若改了安装方式：同步 `README.md` / `README.zh.md` / `INSTALL.md` / `skills/*/README*` / `skills/bailian-protocol/assets/setup.md` 中的 `npx skills add …` 示例（改 `INSTALL.md` 时按 [install-doc-change.md](install-doc-change.md) 同步静态页）
+
+### C. 归属与生成
+
+- [ ] 新一级命令组归属领域时：改 `tools/generate-reference.ts` 的 `GROUP_OWNER_SKILL`，并更新**拥有方** skill 的路由表；hub 最多加一行 hand-off
+- [ ] 跑 `pnpm run sync:skill-assets`（或 commit 走 pre-commit），提交生成的 `reference/` 与 version 同步结果
+- [ ] 默认模型若写在领域路由表（如 `bailian-gen`）：与命令 default / [model-add-remove.md](model-add-remove.md) 一并核对
+
+## 完成后自查
+
+```sh
+pnpm run sync:skill-assets
+# 本地试装（测本仓库改动，勿只拉远端）
+npx skills add "$(pwd)" --all -g -y
+```
+
+抽查：打开 `skills/bailian-cli/SKILL.md` 确认无领域子命令明细表、无 `companions`；打开对应领域 skill 确认有「勿猜 flag」与 hand-off。
+
+## 常见漏点
+
+- ✗ hub 路由表再次抄回 image / video / finetune / managed-agent 明细 → token 膨胀且与领域 skill 双份漂移
+- ✗ 重新加回 `companions` 并宣称安装器硬依赖 → 与 Agent Skills / `npx skills` 合同不符
+- ✗ 软 hand-off 写成硬路径 `../bailian-*/SKILL.md` 当执行前提 → 子集安装断链
+- ✗ 只改 SKILL、忘改 `GROUP_OWNER_SKILL` → reference 落错 skill
+- ✗ 手改 `skills/*/reference/*.md` → 下次 generate 被覆盖
+- ✗ 改默认模型只动 flag description / reference，忘改领域 SKILL「When to use which command」表（见 [model-add-remove.md](model-add-remove.md)）
@@ -0,0 +1,165 @@
+# 埋点变更
+
+## 触发条件
+
+- 调整 AEM 命令事件、事件字段或参数 allowlist
+- 调整 `User-Agent`、`x-dashscope-source-config` 或其他后端渠道标识
+- 新增鉴权域、请求网关或绕开统一 Client 的网络出口
+- 排查命令量、成功率、版本、鉴权域或后端渠道数据不一致
+
+## 当前数据流
+
+三套鉴权对应三套请求域，但不代表三套网关使用相同的后端埋点。命令侧另有一套覆盖所有实际执行命令的 AEM 客户端事件，两者必须分开理解。
+
+```text
+命令进入 run
+  ├─ telemetryStage
+  │    ├─ ~/.bailian/telemetry.jsonl
+  │    └─ AEM(pid=bailian-cli-node, event name=命令路径)
+  │
+  └─ authStage
+       ├─ apiKey  → DashScope / 模型域
+       ├─ console → Bailian Console Gateway
+       ├─ openapi → 阿里云 OpenAPI
+       └─ none    → 无凭证域；本地命令也仍有 AEM 命令事件
+```
+
+### 1. 三套鉴权与埋点标识
+
+| 命令声明          | 凭证 / 请求域                                       | 主要请求出口                                                                          | 后端埋点标识                                  | 前端埋点标识（AEM）                              |
+| ----------------- | --------------------------------------------------- | ------------------------------------------------------------------------------------- | --------------------------------------------- | ------------------------------------------------ |
+| `auth: "apiKey"`  | API Key；DashScope / OpenAI-compatible 模型域       | `Client.request/requestJson`、`McpClient`、Managed Agent instrumented fetch、上传策略 | 有：`User-Agent`、`x-dashscope-source-config` | 有：`pid=bailian-cli-node`、`authMethod=apiKey`  |
+| `auth: "console"` | Console access token；Bailian Console Gateway       | `callConsoleGateway()` → `/cli/api.json`                                              | 无                                            | 有：`pid=bailian-cli-node`、`authMethod=console` |
+| `auth: "openapi"` | AccessKey ID/Secret，可选 STS token；阿里云 OpenAPI | `Client.openApiJson()`                                                                | 有：`x-dashscope-source-config`               | 有：`pid=bailian-cli-node`、`authMethod=openapi` |
+| `auth: "none"`    | 无凭证域                                            | 本地逻辑或命令自行管理的登录/配置流程                                                 | 无                                            | 有：`pid=bailian-cli-node`、`authMethod=none`    |
+
+`authMethod` 记录的是命令声明的鉴权域，不是凭证来源。它不会区分 API Key 来自 flag、env 还是 config。
+鉴权域是命令的准入门槛和主请求域，不保证命令内部只有一种网络出口；例如部分 `apiKey` 命令也可能读取匿名 Console 公共目录，Managed Agent 还可能访问其他 provider。
+
+表中的后端埋点按该鉴权域的主要业务请求填写：
+
+- Managed Agent 的 `User-Agent` 对所有 SDK 请求注入；`x-dashscope-source-config` 仅对阿里云 host 注入
+- DashScope 上传策略 `getPolicy` 只有 `x-dashscope-source-config`，没有显式 CLI `User-Agent`
+- OpenAPI 的 ACS 签名头，以及 Console Gateway 的 `product`、`action`、`api` 是鉴权或路由字段，不计为埋点标识
+
+### 2. 后端渠道参数
+
+当前 `x-dashscope-source-config` 结构为：
+
+```json
+{
+  "channel": "bailian-cli",
+  "tags": {
+    "t1": "public",
+    "t2": "bl 或 kscli",
+    "t3": "实际 CLI 版本"
+  }
+}
+```
+
+- `t2` 取产品 `identity.binName`：完整 CLI 为 `bl`，Knowledge Studio CLI 为 `kscli`
+- `t3` 取产品 `identity.version`，由产品入口的 `package.json` 注入
+- `channel` 与 `t1` 是当前固定口径
+- `User-Agent` 是独立标识：`bl` 为 `bailian-cli/<version>`，`kscli` 为 `knowledge-studio-cli/<version>`
+
+source-config 只用于百炼 / DashScope API 侧消费，不发送到通用网络传输：
+
+| 请求                                 | source-config |
+| ------------------------------------ | ------------- |
+| 模型 API、任务提交与轮询             | 有            |
+| Bailian MCP / OpenAPI                | 有            |
+| DashScope 上传策略 `getPolicy`       | 有            |
+| OSS 文件上传                         | 无            |
+| 图片、视频、音频、转录结果下载       | 无            |
+| npm / 二进制更新检查、Skill registry | 无            |
+
+当前已知例外：Pipeline runtime 自建的 `Identity.version` 为 `0.0.0-dev`，因此 Pipeline 内部模型请求的 `t3` 不代表产品包版本；现阶段不纳入本轮收敛。
+
+### 3. 全命令 AEM 客户端埋点
+
+`packages/runtime/src/middleware.ts` 的 `telemetryStage` 包裹 `authStage` 与命令执行，因此成功、业务失败、网络失败和鉴权失败都会形成一次命令事件。事件名是空格连接的命令路径，例如 `text chat`。
+
+以下情况不会形成命令事件，因为没有进入 middleware 的 `run`：
+
+- 根帮助、子命令 `--help`、`--version`
+- 未识别命令、参数解析失败、缺少必填参数
+- `defineCommand.validate` 在 dispatch 阶段拒绝的请求
+
+遥测默认开启；`DO_NOT_TRACK=1` 一票否决，配置文件 `telemetry: false` 也可关闭。关闭后本地和远端均不记录。
+
+单条 `TrackingEvent` 当前包含：
+
+- `command`、`timestamp`、`durationMs`、`success`
+- `cliVersion`、`nodeVersion`、`os`
+- `authMethod`
+- 失败时的 `errorMessage`、`httpStatus`、`requestId`
+- 安全 allowlist 过滤后的 `params`
+
+参数默认不上传，只有 `packages/core/src/telemetry/tracker.ts` 的 `PARAM_ALLOWLIST` 中字段会进入事件。不得加入 prompt、凭证、文件路径、URL、账号/租户/工作空间 ID 或其他用户内容。
+
+事件同时写入两处：
+
+1. 本地 `~/.bailian/telemetry.jsonl`：权限 `0600`，超过 5 MB 后重建
+2. AEM：`pid=bailian-cli-node`，源码运行自动使用 `env=dev`，npm 安装或编译二进制使用 `env=prod`
+
+底层 Node tracker 还会附加公共设备字段：OS 类型/版本、Node 应用名与版本、平台，以及由本机网络标识计算的 MD5 `device_id`。
+
+当前 AEM 事件没有 `binName` 或 `clientName` 产品维度，并且 `bl`、`kscli` 共用 `pid=bailian-cli-node`。两边相同路径的 `config show`、`config set`、`update` 无法仅凭当前事件稳定区分产品；Knowledge 命令虽然因路径映射不同而表现为 `knowledge chat` 与 `chat`，也不应把命令路径当作长期产品标识。后端 source-config 的 `t2` 已能区分 `bl/kscli`，但这个维度尚未进入 AEM 客户端事件。
+
+AEM 映射：
+
+| AEM 字段   | 内容                                      |
+| ---------- | ----------------------------------------- |
+| event name | 命令路径                                  |
+| `et`       | `EXP`                                     |
+| `ext`      | 除 `command`、`params` 外的结构化事件字段 |
+| `c1`       | allowlist 参数                            |
+| `c2`       | `success` / `failure`                     |
+| `c3`       | HTTP status                               |
+| `c4`       | 错误文案，最多 500 字符                   |
+| `c5`       | request ID                                |
+
+远端发送是 best-effort，不得阻塞命令或改变退出码。正常退出最多等待 1 秒，SIGINT 最多等待 500 ms。
+
+## 必查清单
+
+### A. 新增或调整命令
+
+- [ ] `defineCommand({ auth })` 必须声明真实请求域；AEM 的 `authMethod` 直接读取该值
+- [ ] 新命令进入 `run` 后自动有基础事件，不得在命令内重复发送同名事件
+- [ ] 需要按产品分析 AEM 数据时，必须显式设计产品字段；不得从命令路径推断 `bl/kscli`
+- [ ] 只有可枚举、数值或布尔等低风险字段才可加入 `PARAM_ALLOWLIST`
+- [ ] 新增 console raw API flag 时只允许记录公开 API 名，不得记录请求 `data`
+
+### B. 调整后端渠道参数
+
+- [ ] 同时核对 `packages/core/src/client/http.ts`、`mcp.ts`、`instrumented-fetch.ts`、`client.ts` 与 `files/upload.ts`
+- [ ] 产品身份必须来自 `Identity`；不得从命令路径、环境变量或 `process.argv` 猜测
+- [ ] `bl` 与 `kscli` 必须分别验证 `binName`、`clientName`、`version`
+- [ ] OSS、结果文件、npm、二进制和 Skill 下载不得为了业务渠道统计新增 source-config
+- [ ] 改 URL / host 范围时同时执行 [URL / 渠道变更](url-change.md) 清单
+
+### C. 调整 AEM 事件
+
+- [ ] 更新 `TrackingEvent`、`createTrackingEvent()` 与 `buildRemoteAemOptions()` 的字段映射
+- [ ] 本地 JSONL 与远端 AEM 必须基于同一结构化事件，不能维护两套字段口径
+- [ ] 成功与失败均覆盖；遥测异常必须静默且不改变业务退出码
+- [ ] 检查 `DO_NOT_TRACK=1` 与 `telemetry: false` 两个关闭入口
+- [ ] 错误字段不得额外拼接 token、请求体、prompt 或本地路径
+
+## 完成后自查
+
+```sh
+rg -n "trackingHeaders|x-dashscope-source-config|User-Agent" packages --glob '*.ts'
+rg -n "trackCommandExecution|PARAM_ALLOWLIST|buildRemoteAemOptions" packages/core packages/runtime --glob '*.ts'
+vp check
+vp test packages/core/tests packages/commands/tests/e2e/auth.e2e.test.ts
+```
+
+## 常见漏点
+
+- ✗ 只看 AEM 命令事件，误以为它能替代网关侧请求渠道统计
+- ✗ 把 `authMethod` 当成实际凭证来源；它只是命令声明的鉴权域
+- ✗ 新增 bypass `fetch` 后漏掉应由网关消费的 source-config，或把它发给 OSS / npm / 第三方下载地址
+- ✗ 只改 `bl` 入口，导致 `kscli` 的产品名或版本标签错误
+- ✗ 把帮助、版本或参数校验失败算进“全部命令”；这些路径当前没有进入 telemetry middleware
@@ -51,7 +51,7 @@ grep -rnE "https://dashscope[a-z-]*\.aliyuncs\.com" packages/ --include="*.ts" \

 ### B. 非 TS 文件(只能人工同步,无法 import)

- [ ] `skills/bailian-cli/reference/` 各 `<group>.md` 中 API/控制台 URL(`generate:reference` 重建后核对并提交)
+- [ ] `skills/*/reference/` 各 `<group>.md` 中 API/控制台 URL(`generate:reference` 重建后核对并提交)
 - [ ] `README.md` / `README.zh.md` 中所有 URL

 ### C. 渠道追踪参数
@@ -25,6 +25,7 @@
    "wiki:crawl": "node tools/wiki-crawler/index.mjs",
    "test:stress": "node packages/cli/tests/stress/run.mjs"
  },
+  "dependencies": {},
  "devDependencies": {
    "tsx": "catalog:",
    "vite-plus": "catalog:"
@@ -13,8 +13,9 @@

 ---

-_Chat with Qwen, generate images & videos, understand images, call agents,_
-_manage memory, search the web — all from your terminal._
+_Chat with Qwen, generate and edit images and videos, understand images, synthesize_
+_and recognize speech, call apps, manage memory, retrieve knowledge, search the web —_
+_every AI capability, one command away._

 _Built for AI Agents. Every command works as a structured tool call._

@@ -22,28 +23,16 @@ _Built for AI Agents. Every command works as a structured tool call._

 ## Features

-Equip your AI Agent out-of-the-box with these capabilities, composable across complex tasks:
+- **Model generation** — Full-modality generation across text, image, video, and speech, with editing and reference-based generation
+- **Asset understanding** — Parse and ask questions about images, documents, audio, and long videos
+- **App orchestration** — Call Managed Agents, agents, and workflows published on Aliyun Model Studio, wired to knowledge bases, memory, web search, and MCP tools
+- **Training & deployment** — Validate and upload datasets, fine-tune models, deploy dedicated models as endpoints
+- **Account operations** — Login, UI-based configuration, model marketplace, usage and quota, rate-limit increases, team seat management
+- **Plan onboarding** — Connect subscription plans such as Token Plan to the CLI and common coding agents in one step

- **Text chat** — Qwen3.7-max: major gains in agentic coding, frontend coding, and vibe coding
- **Multimodal (Omni)** — Full omni-modal support across text + image + audio + video
- **Image generation & editing** — Qwen-Image 2.0: pro text rendering, photorealism, strong semantic adherence, multi-image composition
- **Video generation & editing** — happyhorse-1.1 series: text-/image-/reference-to-video and natural-language video editing (up to 9-image reference)
- **Speech synthesis & recognition** — CosyVoice streaming TTS, voice cloning from 5–20s samples; FunAudio-ASR covers 30 languages including 7 Chinese dialects and 20+ Mandarin accents
- **Image & video understanding** — Qwen-VL: long-form video analysis, chart/document parsing, visual reasoning, multilingual OCR
- **Coding agent setup** — Configure Claude Code, Qwen Code, OpenCode, OpenClaw, Hermes Agent, or Codex to use DashScope with `bl config agent`
+> **Note:** App orchestration, training & deployment, account operations, and plan onboarding are currently available only to China site (aliyun.com) account holders and are not yet supported for international / global site accounts.

-> **Note:** The features below are currently available only to China site (aliyun.com) account holders and are not yet supported for international / global site accounts.
-
- **Knowledge base & memory** — Multimodal RAG retrieval and cross-session memory for personalized, coherent dialogue
- **App calls** — Invoke agents and workflows already published on Aliyun Model Studio
- **MCP integration** — Orchestrate Bailian MCP servers: list services, inspect tools, and invoke any tool directly from the terminal
- **Web search** — Real-time internet retrieval for up-to-date, accurate answers
- **Model recommendation** — Describe your scenario and get best-fit model suggestions; supports scoped search, model comparison, and alternative discovery
- **Fine-tuning & deployment** — Upload datasets, create text/audio/image fine-tune jobs (`finetune text|audio|image create`; text covers SFT/LoRA/DPO/CPT), probe job status non-blockingly (`finetune watch`), query per-model training capability (`finetune capability`), and deploy trained models as endpoints (`deploy text|audio|image create`)
- **Console capabilities** — Browse the model marketplace (`model list`) and Bailian apps (`app list`), review a unified usage view (`usage summary`), check free-tier quota (`usage free`), view model usage statistics (`usage stats`), manage workspaces (`workspace list`), and manage rate limits (`quota list/request/check/history`)
- **Local file auto-upload** — Every URL parameter accepts a local path; uploaded to free temp storage with 48-hour validity
-
-## Showcase: One-Sentence Cinematic Video
+## Showcase 1: A Cinematic Short Film from One Sentence

 <p align="center">
  <a href="https://cloud.video.taobao.com/vod/dS2F4huqbw5Nfe5L3wwb3grz2q2DNYD3retq8dU-iHo.mp4">
@@ -56,26 +45,45 @@ Equip your AI Agent out-of-the-box with these capabilities, composable across co
 A complete **2-minute, 16:9 cinematic short film** — produced end-to-end from a single natural-language sentence, with **zero manual editing**. This showcase demonstrates how an AI Agent can compose a multi-step creative pipeline by orchestrating three primitives:

 - **[Qwen Code](https://github.com/QwenLM/qwen-code)** — the agentic coding model that interprets the user's intent and drives the workflow
- **[Aliyun Model Studio CLI](https://bailian.console.aliyun.com/cli?source_channel=cli_github&)** — invokes **HappyHorse 1.1**, Aliyun Model Studio's text-/image-/reference-to-video generation model
+- **[Aliyun Model Studio CLI](https://github.com/modelstudioai/cli/)** — invokes **HappyHorse 1.1**, Aliyun Model Studio's text-/image-/reference-to-video generation model
 - **[spark-video Skill](https://github.com/JohnKeating1997/spark-video)** — handles scene decomposition, storyboarding, shot continuity, and final stitching

 ### The single prompt

 > _"Generate a roughly 2-minute video in Japanese cinematic style — a sweet, innocent first-love story about a high-school girl. The plot should be heart-fluttering enough to make viewers want to fall in love. Aspect ratio: 16:9."_
->
-> _(Original: "帮我生成一段日系影视风格，高中女生的青涩初恋故事，剧情高甜，让人看了想谈恋爱，2分钟左右的视频，尺寸是16:9")_

-### How it works
+## Showcase 2: A Short-Film Director Managed Agent from One Sentence

-1. **Qwen Code** parses the request, plans the narrative beats, and decides which tools to call.
-2. The **spark-video Skill** breaks the story into shots, writes per-shot prompts, and enforces visual continuity (characters, lighting, palette, lens language).
-3. **`bl video generate`** dispatches each shot to **HappyHorse 1.1** in parallel.
-4. The skill stitches all clips back together into a single 16:9 / ~2-min deliverable.
+<p align="center">
+  <a href="https://cloud.video.taobao.com/vod/2v0GYLbJSQb2saj4iopTJDW3iRIHsintYlK-wTKbhqE.mp4">
+    <img src="https://img.alicdn.com/imgextra/i4/6000000001674/O1CN01xhzixhxltbH3LxWu_!!6000000001674-0-tbvideo.jpg" alt="Click to play the demo video" width="720" />
+  </a>
+</p>

-No timeline scrubbing. No frame-by-frame editing. Just one sentence → one video.
+<p align="center"><i>👆 Click the cover to play the full demo</i></p>
+
+One sentence builds a reusable cloud-side short-film director for storyboarding, storyboard image generation, and video creation:
+
+- **[Qwen Code](https://github.com/QwenLM/qwen-code)** — understands the requirement and generates the agent configuration
+- **[Aliyun Model Studio CLI](https://github.com/modelstudioai/cli/)** — validates the configuration, previews the changes, and completes the deployment
+- **[Managed Agent](https://bailian.console.aliyun.com/cn-beijing/?tab=managed-agents#/managed-agents/quick-start)** — runs the director role along with its skills and tools in the cloud
+
+### The single prompt
+
+> _"Build me a Managed Agent app that can produce short films — a director expert that generates videos and can also design the matching storyboards."_

 ## Installation

+**Agent install (recommended)**
+
+Send the following to your Agent — it will detect your environment, then install and verify the CLI for you:
+
+```text
+Please read https://bailian.aliyun.com/cli/install.md and install the Aliyun Model Studio CLI for me
+```
+
+**Manual install (npm)**
+
 ```bash
 npm install -g bailian-cli
 npx skills add modelstudioai/cli --all -g
@@ -85,91 +93,29 @@ npx skills add modelstudioai/cli --all -g

 ## Quick Start

-```bash
-# Authenticate, recommended
-bl auth login --console
+Once installed, just describe your task to your AI Agent — no need to assemble commands by hand.

-# Or authenticate with an API key
-bl auth login --api-key sk-xxxxx
-
-# Or use Token Plan (Base URL built in; the key is tested during login)
-bl auth login --config token-plan --api-key sk-sp-xxxxx
-
-# Configure a coding agent to use DashScope
-bl config agent --agent codex --base-url https://dashscope.aliyuncs.com/compatible-mode/v1 --api-key sk-xxxxx --model qwen3-coder-plus
-
-# Chat with Qwen
-bl text chat --message "What is DashScope?"
-
-# Multimodal chat (text + image + audio + video)
-bl omni --message "Describe this image" --image ./photo.jpg
-
-# Generate an image
-bl image generate --prompt "A cat in a spacesuit" --out-dir ./images/
-
-# Generate a video from local image
-bl video generate --image ./cat.png --prompt "Make the cat move" --download cat.mp4
-
-# Model recommendation — find the best model for your use case
-bl advisor recommend --message "I need a visual-understanding chatbot"
-
-# Compare specific models
-bl advisor recommend --message "qwen-max vs deepseek-v3 for code generation"
-
-# Browser login (required for console capability commands)
-bl auth login --console
-
-# Fine-tune & deploy — a one-shot train-to-serve workflow
-bl dataset upload --file ./train.jsonl                 # Upload a .jsonl dataset (validated first)
-bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft-lora  # Local paths auto-upload
-bl finetune watch --job-id ft-xxx --output json       # Non-blocking probe (running/succeeded return 0; failed/canceled report an error)
-bl finetune capability --model qwen3-8b               # Which training types a model supports
-bl deploy text create --model qwen3-8b --name my-svc --plan mu  # Deploy the trained model as an endpoint
-
-# Browse models / apps / free-tier quota / usage statistics / workspaces
-bl model list                                        # Browse model families and pricing
-bl app list
-bl usage summary                                     # Unified view: free-tier quota + recent usage overview
-bl usage free                                         # Free-tier quota across models (add --model/--expiring/--sort)
-bl usage stats --workspace-id <id>                    # Model usage statistics (add --model for per-model)
-bl workspace list                                     # List all workspaces
-
-# Rate limit management (list / check / request / history)
-bl quota list                                         # View RPM/TPM limits (add --model to filter)
-bl quota check                                        # Current usage vs rate limits (add --model/--period)
-bl quota request --model qwen3.6-plus --tpm 6000000   # Request a temporary TPM increase
-bl quota history                                      # View quota-change history
-
-# Token Plan team management (requires AK/SK, see auth below)
-bl token-plan list-seats                                # View subscription seat details
-bl token-plan add-member --account-name dev --org-id org_xxx
-bl token-plan assign-seats --workspace-id ws_xxx --seat-type standard --account-id acc_xxx
-bl token-plan create-key --account-id acc_xxx --workspace-id ws_xxx
-```
+| Scenario                 | What to say to your Agent                                                         |
+| ------------------------ | --------------------------------------------------------------------------------- |
+| Managed Agent            | "Create a Managed Agent that can generate short-film storyboards and videos."     |
+| Image & video generation | "Generate an image of a cat in a spacesuit on Mars, then turn it into a video."   |
+| Usage & quota            | "Show my recent model usage, free-tier quota, and rate limits."                   |
+| Model selection          | "Recommend a model for image understanding and customer support."                 |
+| About Bailian CLI        | "Tell me what Bailian CLI can do for me, and suggest how to use it for my needs." |

 > More examples and scenarios: [Aliyun Model Studio CLI Site](https://bailian.console.aliyun.com/cli?source_channel=cli_github&)

 ## Authentication

-### DashScope API Key
+### API Key

 Required for most commands. Get your key from the [DashScope Console](https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key).

 ```bash
-# Option 1: Environment variable
-export DASHSCOPE_API_KEY=sk-xxxxx
-
-# Option 2: Login command (persisted to ~/.bailian/config.json)
 bl auth login --api-key sk-xxxxx
-
-# Option 3: Per-command flag
-bl text chat --api-key sk-xxxxx --message "Hello"
 ```

-### Token Plan API Key
-
-Get or copy the API key from the [Token Plan subscription overview](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview).
-The CLI has the default Token Plan Base URL built in. Login tests the key first, then saves and activates the `token-plan` config only when validation succeeds.
+Get or copy your Token Plan API key from the [Token Plan subscription overview](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview).

 ```bash
 bl auth login --config token-plan --api-key sk-sp-xxxxx
@@ -177,26 +123,20 @@ bl auth login --config token-plan --api-key sk-sp-xxxxx

 ### Console Login (OAuth)

-Required for console capability commands (`model list`, `app list`, `usage summary/free/stats`, `workspace list`, `quota list/request/check/history`). Opens the Bailian console in your browser to sign in.
+Required for console capability commands (model list, app list, MCP list, workspace, usage queries, rate-limit increases, direct console calls). Opens the Bailian console in your browser to sign in.

 ```bash
 bl auth login --console
 ```

-### Alibaba Cloud OpenAPI AK/SK (Token Plan only)
+### Alibaba Cloud OpenAPI AK/SK

-Required for the `token-plan` command group. Get your AccessKey from [RAM Console](https://ram.console.aliyun.com/manage/ak).
+Token Plan seat and member management requires an Alibaba Cloud AccessKey. Get yours from the [RAM Console](https://ram.console.aliyun.com/manage/ak).

 > Recommended: create a RAM sub-account with minimum privileges instead of using the root account's AK/SK.

 ```bash
-# Option 1: Login command (persisted to ~/.bailian/config.json)
 bl auth login --open-api --access-key-id LTAI5t... --access-key-secret ...
-
-# Option 2: Environment variables
-export ALIBABA_CLOUD_ACCESS_KEY_ID=LTAI5t...
-export ALIBABA_CLOUD_ACCESS_KEY_SECRET=...
-export BAILIAN_WORKSPACE_ID=ws-...
 ```

 ## Configuration
@@ -205,17 +145,31 @@ export BAILIAN_WORKSPACE_ID=ws-...
 # View current config
 bl config show

-# Set defaults
-bl config set --key base_url --value https://dashscope-us.aliyuncs.com
-bl config set --key default_text_model --value qwen-turbo
-bl config set --key timeout --value 600
+# List all config profiles
+bl config list

-# Self-update to latest version
-bl update
+# Switch config profile
+bl config use --name token-plan
 ```

 Config file location: `~/.bailian/config.json`

+## Update
+
+```bash
+bl update
+```
+
+Upgrades the CLI to the latest version and refreshes the installed Agent Skills. Release notes for every version live in [CHANGELOG.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG.md).
+
+## Contributing
+
+Bug reports, feature requests, and PRs are welcome. See [CONTRIBUTING.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING.md) for developer setup, repo layout, and the workflow for adding or changing commands.
+
+Scan the QR code to join the Aliyun Model Studio CLI DingTalk user group for usage help, troubleshooting, bug reports, and tips from other users.
+
+<img src="https://img.alicdn.com/imgextra/i3/O1CN015uuhYGb6j0L12xJZ_!!6000000006304-2-tps-516-485.png" alt="Aliyun Model Studio CLI DingTalk user group" width="240" />
+
 ## Links

 | Resource                     | URL                                                                                       |
@@ -227,11 +181,3 @@ Config file location: `~/.bailian/config.json`
 | Get API Key                  | https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key |
 | Get Token Plan API Key       | https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview         |
 | Get AccessKey                | https://ram.console.aliyun.com/manage/ak                                                  |
-
-## Changelog
-
-Release notes for every version live in [CHANGELOG.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG.md).
-
-## Contributing
-
-Bug reports, feature requests, and PRs are welcome. See [CONTRIBUTING.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING.md) for developer setup, repo layout, and the workflow for adding or changing commands.
@@ -22,28 +22,16 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_

 ## 功能特性

-让您的 AI Agent 开箱即具备以下能力，并可在复杂任务中自动组合调用：
+- **模型生成** — 文本、图像、视频、语音全模态生成，支持编辑与参考生成
+- **素材理解** — 图像、文档、音频、长视频的解析与问答
+- **应用编排** — 调用百炼已发布的 Managed Agent、智能体和工作流，接入知识库、记忆库、联网搜索与 MCP 工具
+- **模型训推** — 数据集校验上传、模型精调、专属模型部署上线
+- **账号运维** — 授权登录、界面化配置、模型市场、用量与额度、限流提额、团队席位管理
+- **套餐接入** — 支持 Token Plan 等订阅计划一键接到 CLI 和常见 Coding Agent

- **文本对话** — Qwen3.7-max：Agentic coding、前端编程、Vibe coding 等能力显著增强
- **全模态对话** — 文本 + 图像 + 音频 + 视频全模态支持
- **图像生成与编辑** — Qwen-Image 2.0：专业文字渲染、真实质感、强语义遵循、多图合成
- **视频生成与编辑** — happyhorse-1.1 系列，支持文生 / 图生 / 参考生（最多 9 张图参考）/ 自然语言视频编辑
- **语音合成与识别** — CosyVoice 实时流式合成，5-20s 样本即可克隆；FunAudio-ASR 覆盖 30 种语种，含汉语七大方言与 20+ 口音官话
- **图像与视频理解** — Qwen-VL：长视频解析、复杂图表与文档识别、视觉推理、多语种 OCR
- **Coding Agent 配置** — 使用 `bl config agent` 将 Claude Code、Qwen Code、OpenCode、OpenClaw、Hermes Agent 或 Codex 配置为使用 DashScope
+> **注意：** 应用编排、模型训推、账号运维和套餐接入目前仅支持中国站（aliyun.com）账号，暂不支持国际站 / 全球站账号。

-> **注意：** 以下功能目前仅对中国站（aliyun.com）账号开放，国际站 / 全球站账号暂不支持。
-
- **知识库与记忆库** — 多模态 RAG 检索 + 跨会话记忆，提供个性化连贯对话体验
- **应用调用** — 调用已发布在阿里云百炼平台上的智能体与工作流应用
- **MCP 集成** — 统一调度百炼 MCP 服务：列出服务、查看工具、直接在终端调用任意工具
- **联网搜索** — 实时互联网信息检索，提升回答准确性及时效性
- **模型推荐** — 描述你的场景，智能推荐最适合的模型；支持限定范围搜索、模型对比和替代发现
- **微调与部署** — 上传数据集、创建文本/音频/图像调优任务（`finetune text|audio|image create`；文本涵盖 SFT/LoRA/DPO/CPT）、非阻塞探测任务状态（`finetune watch`）、按模型查训练能力（`finetune capability`），并把训练好的模型部署为推理服务（`deploy text|audio|image create`）
- **控制台能力** — 浏览模型市场（`model list`）和百炼应用（`app list`），查看统一用量视图（`usage summary`），查询模型免费额度（`usage free`），查看模型用量统计（`usage stats`），管理业务空间（`workspace list`），管理限流与提额（`quota list/request/check/history`）
- **本地文件自动上传** — 所有 URL 参数同时支持本地路径，免费临时存储 48 小时
-
-## 示例:一句话生成一部电影短片
+## 示例 1：一句话生成一部电影短片

 <p align="center">
  <a href="https://cloud.video.taobao.com/vod/dS2F4huqbw5Nfe5L3wwb3grz2q2DNYD3retq8dU-iHo.mp4">
@@ -53,27 +41,48 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_

 <p align="center"><i>👆 点击封面播放完整 2 分钟演示</i></p>

-一部完整的 **2 分钟、16:9 电影感短片** —— 由一句自然语言端到端生成,**全程零手动剪辑**。这个示例展示了 AI Agent 如何把三个基础能力编排成一条多步创作流水线:
+一部完整的 **2 分钟、16:9 电影感短片** —— 由一句自然语言端到端生成，**全程零手动剪辑**。这个示例展示了 AI Agent 如何把三个基础能力编排成一条多步创作流水线：

- **[Qwen Code](https://github.com/QwenLM/qwen-code)** —— Agentic coding 模型,解析用户意图、驱动整个工作流
- **[阿里云百炼 CLI](https://github.com/modelstudioai/cli/)** —— 调用 **HappyHorse 1.1**,百炼的文生/图生/参考生视频模型
+- **[Qwen Code](https://github.com/QwenLM/qwen-code)** —— Agentic coding 模型，解析用户意图、驱动整个工作流
+- **[阿里云百炼 CLI](https://github.com/modelstudioai/cli/)** —— 调用 **HappyHorse 1.1**，百炼的文生/图生/参考生视频模型
 - **[spark-video Skill](https://github.com/JohnKeating1997/spark-video)** —— 负责场景拆分、分镜设计、镜头连贯性和最终拼接

 ### 唯一的提示词

-> _"帮我生成一段日系影视风格,高中女生的青涩初恋故事,剧情高甜,让人看了想谈恋爱,2 分钟左右的视频,尺寸是 16:9"_
+> _“帮我生成一段日系影视风格，高中女生的青涩初恋故事，剧情高甜，让人看了想谈恋爱，2 分钟左右的视频，尺寸是 16:9。”_

-### 工作流程
+## 示例 2：一句话构建短片导演 Managed Agent

-1. **Qwen Code** 解析需求、规划叙事节奏,决定要调用哪些工具。
-2. **spark-video Skill** 把故事拆成镜头、为每个镜头写提示词,并保证视觉连贯性(角色、光线、色调、镜头语言)。
-3. **`bl video generate`** 把每个镜头并行下发给 **HappyHorse 1.1**。
-4. Skill 把所有片段拼成最终的 16:9 / 约 2 分钟成片。
+<p align="center">
+  <a href="https://cloud.video.taobao.com/vod/2v0GYLbJSQb2saj4iopTJDW3iRIHsintYlK-wTKbhqE.mp4">
+    <img src="https://img.alicdn.com/imgextra/i4/6000000001674/O1CN01xhzixhxltbH3LxWu_!!6000000001674-0-tbvideo.jpg" alt="点击播放演示视频" width="720" />
+  </a>
+</p>

-没有时间线拖拽,没有逐帧剪辑。一句话 → 一部短片。
+<p align="center"><i>👆 点击封面播放完整演示</i></p>
+
+一句话构建一个可复用的云端短片导演，用于分镜设计、分镜图生成和视频创作：
+
+- **[Qwen Code](https://github.com/QwenLM/qwen-code)** —— 理解需求并生成 Agent 配置
+- **[阿里云百炼 CLI](https://github.com/modelstudioai/cli/)** —— 校验配置、预览变更并完成部署
+- **[Managed Agent](https://bailian.console.aliyun.com/cn-beijing/?tab=managed-agents#/managed-agents/quick-start)** —— 在云端运行导演角色及其 Skill 和工具
+
+### 唯一的提示词
+
+> _“帮我构建一个 managedagent 应用，能够实现短片拍摄，导演专家生成视频，然后也能进行设计对应的分镜图。”_

 ## 安装

+**Agent 安装（推荐）**
+
+把下面这句话发给你的 Agent，它会自行判断环境并完成安装与校验：
+
+```text
+请阅读：https://bailian.aliyun.com/cli/install.md 并按照说明为我安装阿里云百炼 CLI
+```
+
+**手动安装（npm）**
+
 ```bash
 npm install -g bailian-cli
 npx skills add modelstudioai/cli --all -g
@@ -83,91 +92,29 @@ npx skills add modelstudioai/cli --all -g

 ## 快速开始

-```bash
-# 认证（推荐浏览器登录）
-bl auth login --console
+安装完成后，直接在 AI Agent 中描述你的任务，无需手动拼接命令。

-# 或使用 API key 认证
-bl auth login --api-key sk-xxxxx
-
-# 或使用 Token Plan（已内置 Base URL，登录时自动测试 Key）
-bl auth login --config token-plan --api-key sk-sp-xxxxx
-
-# 配置 Coding Agent 使用 DashScope
-bl config agent --agent codex --base-url https://dashscope.aliyuncs.com/compatible-mode/v1 --api-key sk-xxxxx --model qwen3-coder-plus
-
-# 和通义千问对话
-bl text chat --message "你好，介绍一下阿里云百炼平台"
-
-# 多模态对话（文本 + 图片 + 音频 + 视频）
-bl omni --message "描述这张图片" --image ./photo.jpg
-
-# 生成图片
-bl image generate --prompt "一只穿太空服的猫在火星上" --out-dir ./images/
-
-# 图生视频（本地文件自动上传）
-bl video generate --image ./cat.png --prompt "让画面中的猫动起来" --download cat.mp4
-
-# 模型推荐 — 根据场景推荐最适合的模型
-bl advisor recommend --message "我要做一个能理解图片的客服机器人"
-
-# 对比特定模型
-bl advisor recommend --message "qwen-max 和 deepseek-v3 哪个更适合做代码生成"
-
-# 浏览器登录（控制台能力相关命令需要）
-bl auth login --console
-
-# 微调与部署 — 从训练到服务的一站式流程
-bl dataset upload --file ./train.jsonl                 # 上传 .jsonl 数据集（先校验）
-bl finetune text create --model qwen3-8b --datasets ./train.jsonl --training-type sft-lora  # 本地路径自动上传
-bl finetune watch --job-id ft-xxx --output json       # 非阻塞探测（运行中/成功返回 0；失败/取消报错）
-bl finetune capability --model qwen3-8b               # 查询模型支持哪些训练方式
-bl deploy text create --model qwen3-8b --name my-svc --plan mu  # 把训练好的模型部署为推理服务
-
-# 浏览模型 / 应用 / 免费额度 / 用量统计 / 业务空间
-bl model list                                        # 浏览模型系列与价格信息
-bl app list
-bl usage summary                                     # 统一视图：免费额度 + 近期用量概览
-bl usage free                                         # 各模型免费额度（可加 --model/--expiring/--sort）
-bl usage stats --workspace-id <id>                    # 模型用量统计（加 --model 查单模型）
-bl workspace list                                     # 列出所有业务空间
-
-# 限流管理与提额（list / check / request / history）
-bl quota list                                         # 查看 RPM/TPM 限额（加 --model 过滤）
-bl quota check                                        # 当前用量 vs 限流阈值（加 --model/--period）
-bl quota request --model qwen3.6-plus --tpm 6000000   # 申请临时 TPM 提额
-bl quota history                                      # 查看提额历史记录
-
-# Token Plan 团队版管理（需 AK/SK，见下方认证说明）
-bl token-plan list-seats                                # 查看订阅席位明细
-bl token-plan add-member --account-name dev --org-id org_xxx
-bl token-plan assign-seats --workspace-id ws_xxx --seat-type standard --account-id acc_xxx
-bl token-plan create-key --account-id acc_xxx --workspace-id ws_xxx
-```
+| 场景             | 可以这样对 Agent 说                                                     |
+| ---------------- | ----------------------------------------------------------------------- |
+| Managed Agent    | “帮我创建一个能够生成短片分镜和视频的 Managed Agent。”                  |
+| 图片和视频生成   | “生成一张穿着太空服的猫站在火星上的图片，再把它制作成一段视频。”        |
+| 用量与额度       | “查看最近的模型用量、免费额度和限流情况。”                              |
+| 模型选型         | “推荐一个适合图片理解和智能客服的模型。”                                |
+| 了解 Bailian CLI | “介绍一下 Bailian CLI 能帮我完成哪些任务，并根据我的需求推荐使用方式。” |

 > 更多案例与使用场景：[阿里云百炼 CLI 官方主页](https://bailian.console.aliyun.com/cli?source_channel=cli_github&)

 ## 认证方式

-### DashScope API Key
+### API Key

 大部分命令均需要 API Key。前往 [DashScope 控制台](https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key) 获取。

 ```bash
-# 方式一：环境变量
-export DASHSCOPE_API_KEY=sk-xxxxx
-
-# 方式二：登录命令（持久化到 ~/.bailian/config.json）
 bl auth login --api-key sk-xxxxx
-
-# 方式三：命令行参数
-bl text chat --api-key sk-xxxxx --message "你好"
 ```

-### Token Plan API Key
-
-前往 [Token Plan 订阅详情](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) 获取或复制 API Key。
-CLI 已内置 Token Plan 的默认 Base URL；登录命令会先测试 Key，通过后才保存并激活 `token-plan` 配置。
+Token Plan 的 API Key 前往 [Token Plan 订阅详情](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) 获取或复制。

 ```bash
 bl auth login --config token-plan --api-key sk-sp-xxxxx
@@ -175,26 +122,20 @@ bl auth login --config token-plan --api-key sk-sp-xxxxx

 ### 控制台登录（OAuth）

-控制台能力命令（`model list`、`app list`、`usage summary/free/stats`、`workspace list`、`quota list/request/check/history`）需要使用此登录方式。打开浏览器跳转百炼控制台完成登录。
+控制台能力命令（模型列表、应用列表、MCP 列表、工作空间、用量查询、限流提额、控制台直调）需要使用此登录方式。打开浏览器跳转百炼控制台完成登录。

 ```bash
 bl auth login --console
 ```

-### 阿里云 OpenAPI AK/SK（仅 Token Plan）
+### 阿里云 OpenAPI AK/SK

-`token-plan` 命令组需要阿里云 AccessKey。前往 [RAM 控制台](https://ram.console.aliyun.com/manage/ak) 获取。
+Token Plan 的席位与成员管理需要阿里云 AccessKey。前往 [RAM 控制台](https://ram.console.aliyun.com/manage/ak) 获取。

 > 建议：创建 RAM 子账号并授予最小权限，避免使用主账号 AK/SK。

 ```bash
-# 方式一：登录命令（持久化到 ~/.bailian/config.json）
 bl auth login --open-api --access-key-id LTAI5t... --access-key-secret ...
-
-# 方式二：环境变量
-export ALIBABA_CLOUD_ACCESS_KEY_ID=LTAI5t...
-export ALIBABA_CLOUD_ACCESS_KEY_SECRET=...
-export BAILIAN_WORKSPACE_ID=ws-...
 ```

 ## 配置
@@ -203,17 +144,31 @@ export BAILIAN_WORKSPACE_ID=ws-...
 # 查看当前配置
 bl config show

-# 设置默认值
-bl config set --key base_url --value https://dashscope-us.aliyuncs.com
-bl config set --key default_text_model --value qwen-turbo
-bl config set --key timeout --value 600
+# 查看全部配置档
+bl config list

-# 自更新到最新版本
-bl update
+# 切换配置档
+bl config use --name token-plan
 ```

 配置文件位置：`~/.bailian/config.json`

+## 更新
+
+```bash
+bl update
+```
+
+升级 CLI 至最新版本，并同步更新已安装的 Agent Skills。每个版本的变更详情记录在 [CHANGELOG.zh.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG.zh.md)。
+
+## 参与贡献
+
+欢迎提 Issue、Feature Request 和 PR。开发环境搭建、仓库结构、新增/修改命令的工作流请见 [CONTRIBUTING.zh.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING.zh.md)。
+
+欢迎扫码加入阿里云百炼 CLI 钉钉用户交流群，获取使用答疑、问题排查、Bug 反馈和使用经验交流支持。
+
+<img src="https://img.alicdn.com/imgextra/i3/O1CN015uuhYGb6j0L12xJZ_!!6000000006304-2-tps-516-485.png" alt="阿里云百炼 CLI 钉钉用户交流群" width="240" />
+
 ## 相关链接

 | 资源                    | 地址                                                                                      |
@@ -225,11 +180,3 @@ bl update
 | 获取 API Key            | https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key |
 | 获取 Token Plan API Key | https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview         |
 | 获取 AccessKey          | https://ram.console.aliyun.com/manage/ak                                                  |
-
-## 更新日志
-
-每个版本的变更详情记录在 [CHANGELOG.zh.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG.zh.md)。
-
-## 参与贡献
-
-欢迎提 Issue、Feature Request 和 PR。开发环境搭建、仓库结构、新增/修改命令的工作流请见 [CONTRIBUTING.zh.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING.zh.md)。
@@ -1,6 +1,6 @@
 {
  "name": "bailian-cli",
-  "version": "1.13.0",
+  "version": "1.14.2",
  "description": "CLI for Aliyun Model Studio (DashScope) AI Platform.",
  "keywords": [
    "agent",
@@ -25,7 +25,8 @@
  },
  "files": [
    "dist",
-    "README.zh.md"
+    "README.zh.md",
+    "postinstall.js"
  ],
  "type": "module",
  "exports": {
@@ -40,17 +41,19 @@
    "registry": "https://registry.npmjs.org/"
  },
  "scripts": {
-    "generate:reference": "tsx ../../tools/generate-reference.ts && sh -c 'cd ../.. && vp check --fix skills/bailian-cli/reference'",
+    "generate:reference": "tsx ../../tools/generate-reference.ts && sh -c 'cd ../.. && vp check --fix skills/bailian-cli/reference skills/bailian-gen/reference skills/bailian-finetune/reference skills/bailian-managed-agent/reference'",
    "sync:skill-version": "tsx ../../tools/sync-skill-metadata.ts",
    "build": "vp pack",
    "dev": "tsx src/main.ts",
    "test": "vp test",
-    "check": "vp check"
+    "check": "vp check",
+    "postinstall": "node postinstall.js"
  },
  "dependencies": {
    "bailian-cli-commands": "workspace:*",
    "bailian-cli-core": "workspace:*",
-    "bailian-cli-runtime": "workspace:*"
+    "bailian-cli-runtime": "workspace:*",
+    "tar-stream": "catalog:"
  },
  "devDependencies": {
    "@clack/prompts": "^0.7.0",
@@ -0,0 +1,253 @@
+/**
+ * postinstall.js — Wiki data sync (layer 1: triggered by npm install)
+ *
+ * Runs automatically after npm/pnpm installs bailian-cli: unconditionally downloads the full Wiki data
+ * package and overwrites the local directory, ensuring data is in place the first time the user runs
+ * `bl advisor recommend`.
+ *
+ * Flow (unified skill publishing protocol: skills/index.json + one content-addressed object per skill):
+ *   1. Download skills/index.json from public-read OSS, get the bailian-docs-llm-wiki entry
+ *   2. Download skills/bailian-docs-llm-wiki/<entry.object> (sha256-<hex>.tar.br, brotli q6, ~2.3MB);
+ *      legacy fallback to skill.tar.br when the entry has no valid object field
+ *   3. Node built-in brotli decompress + tar-stream extract (per-entry path safety check) to same-volume temp dir,
+ *      then recompute contentHash over the extracted files and reject on mismatch (symmetric with core installer)
+ *   4. renameSync atomic swap into ~/.bailian/skills/bailian-docs-llm-wiki/
+ *   5. Write ~/.bailian/wiki-sync-state.json
+ *   6. Write ~/.bailian/skills/skill-lock.json record (same ledger as bl skill)
+ *
+ * Design constraints:
+ *   - Unconditional overwrite: every install fully replaces, no version comparison
+ *   - Silent failure: any step failure → console.warn → process.exit(0), never blocks install
+ *   - Standalone implementation: does not import bailian-cli-core, avoiding ESM path issues after bundling
+ *   - Depends on Node built-in modules + tar-stream (consistent with sync.ts / publisher skills-publish.mjs)
+ */
+import { createHash } from "node:crypto";
+import {
+  createWriteStream,
+  existsSync,
+  mkdirSync,
+  readdirSync,
+  readFileSync,
+  renameSync,
+  rmSync,
+  writeFileSync,
+} from "node:fs";
+import { homedir } from "node:os";
+import { dirname, join } from "node:path";
+import { Readable } from "node:stream";
+import { pipeline } from "node:stream/promises";
+import { createBrotliDecompress } from "node:zlib";
+import tar from "tar-stream";
+
+const REGISTRY_BASE_URL = "https://bailian-wiki.oss-cn-hangzhou.aliyuncs.com/skills";
+const WIKI_SKILL_NAME = "bailian-docs-llm-wiki";
+const CONFIG_DIR_NAME = ".bailian";
+const SKILL_DIR_NAME = "skills/bailian-docs-llm-wiki";
+const STATE_FILE_NAME = "wiki-sync-state.json";
+const INDEX_KEY = "index.json";
+/** Legacy fixed asset key (entries without a valid content-addressed object field) */
+const LEGACY_ASSET_NAME = "skill.tar.br";
+/** Same strict shape check as core registry.ts: only a valid object name may enter the URL */
+const OBJECT_FILE_RE = /^sha256-[0-9a-f]{64}\.tar\.br$/;
+
+const INDEX_TIMEOUT_MS = 3000;
+const DOWNLOAD_TIMEOUT_MS = 30000;
+
+function getConfigDir() {
+  if (process.env.BAILIAN_CONFIG_DIR) return process.env.BAILIAN_CONFIG_DIR;
+  return join(homedir(), CONFIG_DIR_NAME);
+}
+
+function getCatalogDir() {
+  return join(getConfigDir(), SKILL_DIR_NAME);
+}
+
+function getStatePath() {
+  return join(getConfigDir(), STATE_FILE_NAME);
+}
+
+function getSkillLockPath() {
+  return join(getConfigDir(), "skills", "skill-lock.json");
+}
+
+/**
+ * Record this sync in skill-lock.json (same ledger as bl skill; list shows installed).
+ * Semantics aligned with upsertSkillLockEntry in core/src/skills/lock.ts: shallow-merge with the existing
+ * entry, preserving fields like links written by bl skill add; rebuild as empty table if lock is corrupted/unrecognized.
+ * best-effort: failure does not affect data sync results.
+ */
+function upsertSkillLock(name, entry) {
+  try {
+    let lock = { version: 1, skills: {} };
+    try {
+      const parsed = JSON.parse(readFileSync(getSkillLockPath(), "utf-8"));
+      if (parsed?.version === 1 && parsed.skills && typeof parsed.skills === "object") {
+        lock = parsed;
+      }
+    } catch {
+      /* absent/corrupted → empty table */
+    }
+    lock.skills[name] = { ...lock.skills[name], ...entry };
+    mkdirSync(dirname(getSkillLockPath()), { recursive: true });
+    writeFileSync(getSkillLockPath(), JSON.stringify(lock, null, 2) + "\n");
+  } catch {
+    /* Bookkeeping failure does not block install; advisor-side sync will backfill */
+  }
+}
+
+async function fetchJson(url, timeoutMs) {
+  const res = await fetch(url, { signal: AbortSignal.timeout(timeoutMs) });
+  if (!res.ok) throw new Error(`HTTP ${res.status}`);
+  return res.json();
+}
+
+async function downloadBuffer(url) {
+  const res = await fetch(url, { signal: AbortSignal.timeout(DOWNLOAD_TIMEOUT_MS) });
+  if (!res.ok) throw new Error(`HTTP ${res.status}`);
+  return Buffer.from(await res.arrayBuffer());
+}
+
+/** tar 条目路径必须是相对路径且不含 ..，防止 tar-slip 逃逸解包目录 */
+function isSafeEntryName(name) {
+  // Symmetric with core skills/extract.ts: backslashes can escape the extraction
+  // dir on Windows (path.join expands "\.." segments, leading "\" hits drive root)
+  if (name.includes("\\") || name.includes("\0")) return false;
+  if (name.startsWith("/") || /^[a-zA-Z]:[\\/]/.test(name)) return false;
+  return !name.split("/").includes("..");
+}
+
+/** Brotli decompress + tar-stream extract into destDir (symmetric with publisher tar.pack()). */
+async function extractTarBr(tarBrBuffer, destDir) {
+  const extract = tar.extract();
+
+  extract.on("entry", (header, stream, next) => {
+    if (!isSafeEntryName(header.name)) {
+      // Same semantics as core skills/extract.ts: destroy so the pipeline rejects with this
+      // error; silence the entry stream to avoid its companion error becoming unhandled
+      stream.on("error", () => {});
+      stream.resume();
+      extract.destroy(new Error(`unsafe tar entry: ${header.name}`));
+      return;
+    }
+    const filePath = join(destDir, header.name);
+    if (header.type === "directory") {
+      mkdirSync(filePath, { recursive: true });
+      stream.resume();
+      stream.on("end", next);
+      return;
+    }
+    mkdirSync(dirname(filePath), { recursive: true });
+    const ws = createWriteStream(filePath);
+    stream.pipe(ws);
+    ws.on("finish", next);
+    ws.on("error", next);
+  });
+
+  await pipeline(Readable.from(tarBrBuffer), createBrotliDecompress(), extract);
+}
+
+/**
+ * Recompute the publisher's deterministic content hash over an extracted directory
+ * (same accumulation as core skills/extract.ts computeDirContentHash): regular files
+ * sorted by "/"-separated relative path, sha256 over relPath + bytes.
+ */
+function computeDirContentHash(dir) {
+  const relPaths = [];
+  const walk = (sub) => {
+    for (const dirent of readdirSync(sub ? join(dir, sub) : dir, { withFileTypes: true })) {
+      const rel = sub ? `${sub}/${dirent.name}` : dirent.name;
+      if (dirent.isDirectory()) walk(rel);
+      else if (dirent.isFile()) relPaths.push(rel);
+    }
+  };
+  walk("");
+  relPaths.sort((left, right) => (left < right ? -1 : left > right ? 1 : 0));
+  const hash = createHash("sha256");
+  for (const rel of relPaths) {
+    hash.update(rel);
+    hash.update(readFileSync(join(dir, rel)));
+  }
+  return `sha256:${hash.digest("hex")}`;
+}
+
+/** Atomic swap: tmpDir (same volume) → catalogDir. */
+function atomicSwap(tmpDir, catalogDir) {
+  mkdirSync(dirname(catalogDir), { recursive: true });
+  const backup = `${catalogDir}.old-${Date.now()}`;
+  if (existsSync(catalogDir)) renameSync(catalogDir, backup);
+  try {
+    renameSync(tmpDir, catalogDir);
+  } catch (err) {
+    if (existsSync(backup) && !existsSync(catalogDir)) renameSync(backup, catalogDir);
+    throw err;
+  }
+  if (existsSync(backup)) rmSync(backup, { recursive: true, force: true });
+}
+
+async function main() {
+  // 1. Download skills/index.json and get the wiki entry
+  const index = await fetchJson(`${REGISTRY_BASE_URL}/${INDEX_KEY}`, INDEX_TIMEOUT_MS);
+  const entry = index?.skills?.[WIKI_SKILL_NAME];
+  if (!entry?.contentHash)
+    throw new Error("no bailian-docs-llm-wiki entry (or contentHash) in index.json");
+
+  // 2. Download the skill archive: content-addressed object first, legacy fixed key as fallback
+  const assetName =
+    entry.object && OBJECT_FILE_RE.test(entry.object) ? entry.object : LEGACY_ASSET_NAME;
+  const tarBuf = await downloadBuffer(`${REGISTRY_BASE_URL}/${WIKI_SKILL_NAME}/${assetName}`);
+
+  // 3. Extract to same-volume temp dir + integrity check + atomic swap
+  const catalogDir = getCatalogDir();
+  const tmpDir = `${catalogDir}.tmp-${process.pid}-${Date.now()}`;
+  try {
+    mkdirSync(tmpDir, { recursive: true });
+    await extractTarBr(tarBuf, tmpDir);
+    // Symmetric with layer 2 (core installer): reject archive/index fingerprint mismatch
+    // before touching the canonical dir
+    if (entry.contentHash.startsWith("sha256:")) {
+      const actualContentHash = computeDirContentHash(tmpDir);
+      if (actualContentHash !== entry.contentHash) {
+        throw new Error(
+          `content hash mismatch: index says ${entry.contentHash}, archive is ${actualContentHash}`,
+        );
+      }
+    }
+    atomicSwap(tmpDir, catalogDir);
+  } catch (err) {
+    if (existsSync(tmpDir)) rmSync(tmpDir, { recursive: true, force: true });
+    throw err;
+  }
+
+  // 4. Write state
+  try {
+    writeFileSync(
+      getStatePath(),
+      JSON.stringify({ lastChecked: Date.now(), contentHash: entry.contentHash }),
+    );
+  } catch {
+    /* state write failure has no impact: first recommend will re-check */
+  }
+
+  // 5. skill-lock.json record: wiki shares the same ledger as bl skill
+  upsertSkillLock(WIKI_SKILL_NAME, {
+    contentHash: entry.contentHash,
+    ...(entry.publishedAt ? { publishedAt: entry.publishedAt } : {}),
+    installedAt: new Date().toISOString(),
+    sourceType: "oss",
+    ...(entry.description ? { description: entry.description } : {}),
+  });
+
+  process.stdout.write(`bailian-cli: wiki data ready (${entry.publishedAt ?? "latest"})\n`);
+}
+
+main().catch((err) => {
+  // Unconditional pass-through: install-time network/permission issues should not block npm install;
+  // sync.ts will fall back to syncing on the first `bl advisor recommend`.
+  const msg = err instanceof Error ? err.message : String(err);
+  process.stderr.write(
+    `bailian-cli: wiki data pre-download skipped (${msg}); will sync automatically on first use.\n`,
+  );
+  // Force a success exit code so a download failure never fails `npm install`.
+  // eslint-disable-next-line unicorn/no-process-exit
+  process.exit(0);
+});
@@ -30,6 +30,10 @@ import {
  memoryDelete,
  memoryProfileCreate,
  memoryProfileGet,
+  memoryProfileList,
+  memoryProfileDetail,
+  memoryProfileUpdate,
+  memoryProfileDelete,
  knowledgeRetrieve,
  knowledgeSearch,
  knowledgeChat,
@@ -89,10 +93,16 @@ import {
  pluginLink,
  pluginList,
  pluginRemove,
+  skillAdd,
+  skillUpdate,
+  skillRemove,
+  skillList,
+  skillInit,
  managedAgentInit,
  managedAgentValidate,
  managedAgentPlan,
  managedAgentApply,
+  managedAgentRun,
  managedAgentDestroy,
  managedAgentStateList,
  managedAgentStateShow,
@@ -144,6 +154,10 @@ export const commands: Record<string, AnyCommand> = {
  "memory delete": memoryDelete,
  "memory profile create": memoryProfileCreate,
  "memory profile get": memoryProfileGet,
+  "memory profile list": memoryProfileList,
+  "memory profile detail": memoryProfileDetail,
+  "memory profile update": memoryProfileUpdate,
+  "memory profile delete": memoryProfileDelete,
  "knowledge retrieve": knowledgeRetrieve,
  "knowledge search": knowledgeSearch,
  "knowledge chat": knowledgeChat,
@@ -203,10 +217,16 @@ export const commands: Record<string, AnyCommand> = {
  "plugin link": pluginLink,
  "plugin list": pluginList,
  "plugin remove": pluginRemove,
+  "skill add": skillAdd,
+  "skill update": skillUpdate,
+  "skill remove": skillRemove,
+  "skill list": skillList,
+  "skill init": skillInit,
  "managed-agent init": managedAgentInit,
  "managed-agent validate": managedAgentValidate,
  "managed-agent plan": managedAgentPlan,
  "managed-agent apply": managedAgentApply,
+  "managed-agent run": managedAgentRun,
  "managed-agent destroy": managedAgentDestroy,
  "managed-agent state list": managedAgentStateList,
  "managed-agent state show": managedAgentStateShow,
@@ -1,6 +1,6 @@
 {
  "name": "bailian-cli-commands",
-  "version": "1.13.0",
+  "version": "1.14.2",
  "description": "Command library for bailian-cli products (knowledge, memory, media, …). See https://www.npmjs.com/package/bailian-cli for usage.",
  "homepage": "https://bailian.console.aliyun.com/cli",
  "bugs": {
@@ -6,6 +6,7 @@ import {
  type GetModelsOptions,
  getModels,
  type IntentProfile,
+  maybeSyncWikiData,
  type PipelineStep,
  type RecommendedModel,
  type RecommendResult,
@@ -248,6 +249,12 @@ export default defineCommand({
    const { settings, flags } = ctx;
    const userInput = flags.message;
    const top = 3;
+
+    // Keep the local wiki catalog fresh: throttled (12h) version check against
+    // the remote manifest, silently replaces data when a newer version exists.
+    // Never throws — a sync failure must not block recommendation.
+    await maybeSyncWikiData();
+
    // Default to JSON for structured output; render boxen cards only when the
    // user explicitly asked for text output.
    const format = settings.outputExplicit ? detectOutputFormat(settings.output) : "json";
@@ -57,7 +57,7 @@ export async function validateAndPersistApiKey(
  const persistBaseUrl = profile.persistBaseUrl
    ? normalizeModelBaseUrl(profile.persistBaseUrl)
    : undefined;
-  const validationModel = "qwen3.7-max";
+  const validationModel = "qwen3.8-max";
  const requestOpts = {
    url: baseUrl + chatPath(),
    method: "POST",
@@ -60,8 +60,8 @@ export const UI_BOOLEAN_KEYS = new Set<string>(["telemetry"]);
 // are surfaced as input placeholders so users can see the effective default
 // without persisting a value that would pin the model.
 export const UI_MODEL_DEFAULTS: Record<string, string> = {
-  default_text_model: "qwen3.7-max",
-  default_image_model: "qwen-image-2.0",
+  default_text_model: "qwen3.8-max",
+  default_image_model: "qwen-image-3.0",
  default_video_model: "happyhorse-1.1-t2v",
  default_speech_model: "cosyvoice-v3-flash",
  default_omni_model: "qwen3.5-omni-plus",
@@ -80,13 +80,14 @@ export interface ModelOption {
 // Only names present in the codebase are listed here — no invented models.
 export const UI_MODEL_CATALOG: Record<string, ModelOption[]> = {
  default_text_model: [
-    { id: "qwen3.7-max", role: "text/chat default" },
+    { id: "qwen3.8-max", role: "text/chat default" },
    { id: "qwen3-coder-plus", role: "coding-oriented (agent config)" },
    { id: "qwen-flash", role: "fast · advisor ranking" },
    { id: "qwen3.6-flash", role: "fast · advisor intent" },
  ],
  default_image_model: [
-    { id: "qwen-image-2.0", role: "image/generate default · sync" },
+    { id: "qwen-image-3.0", role: "image/generate default · sync" },
+    { id: "qwen-image-2.0", role: "image/generate · sync" },
    { id: "qwen-image-max", role: "image/generate · sync" },
    { id: "qwen-image-edit-2.0", role: "image/edit · sync" },
    { id: "wanx2.x", role: "image/generate · async series" },
@@ -1,5 +1,5 @@
 import { defineCommand, detectOutputFormat, deleteDataset, type FlagsDef } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const DELETE_FLAGS = {
  fileId: {
@@ -30,6 +30,7 @@ export default defineCommand({

    if (settings.quiet || format === "text") {
      emitBare(`Deleted ${fileId}.`);
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -1,5 +1,5 @@
 import { defineCommand, detectOutputFormat, getDataset, type FlagsDef } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const GET_FLAGS = {
  fileId: {
@@ -46,7 +46,7 @@ export default defineCommand({
    };

    if (format === "json") {
-      emitResult(item, format);
+      emitResult({ ...item, request_id: response.request_id }, format);
      return;
    }

@@ -58,5 +58,6 @@ export default defineCommand({
    if (item.purpose) emitBare(`purpose:      ${item.purpose}`);
    if (item.created_at) emitBare(`created_at:   ${item.created_at}`);
    if (item.description) emitBare(`description:  ${item.description}`);
+    emitRequestId(response.request_id, settings.quiet);
  },
 });
@@ -1,5 +1,5 @@
 import { defineCommand, detectOutputFormat, listDatasets, type FlagsDef } from "bailian-cli-core";
-import { emitResult, emitBare, formatTable } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime";

 const LIST_FLAGS = {
  page: { type: "number", valueHint: "<n>", description: "Page number (default: 1)" },
@@ -55,7 +55,7 @@ export default defineCommand({
    }));

    if (format === "json") {
-      emitResult({ items, total }, format);
+      emitResult({ items, total, request_id: response.request_id }, format);
      return;
    }

@@ -68,5 +68,6 @@ export default defineCommand({
    const rows = items.map((i) => [i.file_id, i.name, i.size, i.purpose]);
    for (const line of formatTable(headers, rows)) emitBare(line);
    if (total !== undefined) emitBare(`\nTotal: ${total}`);
+    emitRequestId(response.request_id, settings.quiet);
  },
 });
@@ -9,10 +9,9 @@ import {
  MAX_MEDIA_ZIP_BYTES,
  BailianError,
  ExitCode,
-  type DatasetFile,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const UPLOAD_FLAGS = {
  file: {
@@ -135,17 +134,19 @@ export default defineCommand({
      return;
    }

-    const uploaded: DatasetFile = await uploadDataset(ctx.client, {
+    const uploaded = await uploadDataset(ctx.client, {
      filePath,
      purpose,
    });
+    const { request_id, ...file } = uploaded;

    if (settings.quiet) {
-      emitBare(uploaded.file_id);
+      emitBare(file.file_id);
    } else if (format === "text") {
-      emitBare(`Uploaded ${uploaded.name} → file_id=${uploaded.file_id}`);
+      emitBare(`Uploaded ${file.name} → file_id=${file.file_id}`);
+      emitRequestId(request_id, settings.quiet);
    } else {
-      emitResult(uploaded, format);
+      emitResult({ ...file, request_id }, format);
    }
  },
 });
@@ -11,7 +11,7 @@ import {
  type CommandContext,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const CREATE_FLAGS = {
  model: {
@@ -163,6 +163,7 @@ async function runCreate(
    emitBare(
      `\nNext: track readiness with: ${identity.binName} deploy get --deployed-model ${deployment?.deployed_model ?? "<id>"}`,
    );
+    emitRequestId(response.request_id, settings.quiet);
  } else {
    emitResult(response, format);
  }
@@ -7,7 +7,7 @@ import {
  ExitCode,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const DELETE_FLAGS = {
  deployedModel: {
@@ -71,6 +71,7 @@ export default defineCommand({
      emitBare(deployedModel);
    } else if (format === "text") {
      emitBare(`Deleted ${deployedModel}.`);
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -1,5 +1,5 @@
 import { defineCommand, detectOutputFormat, getDeployment, type FlagsDef } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const GET_FLAGS = {
  deployedModel: {
@@ -58,7 +58,7 @@ export default defineCommand({
    if (deployment.gmt_modified) item.updated_at = deployment.gmt_modified;

    if (format === "json") {
-      emitResult(item, format);
+      emitResult({ ...item, request_id: response.request_id }, format);
      return;
    }

@@ -69,5 +69,6 @@ export default defineCommand({
      const display = typeof value === "string" ? value : JSON.stringify(value);
      emitBare(`${label(key)}${display}`);
    }
+    emitRequestId(response.request_id, settings.quiet);
  },
 });
@@ -4,7 +4,7 @@ import {
  listDeployments,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare, formatTable } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime";

 const LIST_FLAGS = {
  page: { type: "number", valueHint: "<n>", description: "Page number (default: 1)" },
@@ -58,7 +58,7 @@ export default defineCommand({
    }));

    if (format === "json") {
-      emitResult({ items, total }, format);
+      emitResult({ items, total, request_id: response.request_id }, format);
      return;
    }

@@ -78,5 +78,6 @@ export default defineCommand({
    ]);
    for (const line of formatTable(headers, rows)) emitBare(line);
    if (total !== undefined) emitBare(`\nTotal: ${total}`);
+    emitRequestId(response.request_id, settings.quiet);
  },
 });
@@ -4,7 +4,7 @@ import {
  listDeployableModels,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare, formatTable } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime";

 const MODELS_FLAGS = {
  page: { type: "number", valueHint: "<n>", description: "Page number (default: 1)" },
@@ -122,7 +122,7 @@ export default defineCommand({
        }
        return out;
      });
-      emitResult({ items, total }, format);
+      emitResult({ items, total, request_id: response.request_id }, format);
      return;
    }

@@ -168,5 +168,6 @@ export default defineCommand({
    ]);
    for (const line of formatTable(headers, rows)) emitBare(line);
    if (total !== undefined) emitBare(`\nTotal: ${total}`);
+    emitRequestId(response.request_id, settings.quiet);
  },
 });
@@ -4,7 +4,7 @@ import {
  scaleDeployment,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const SCALE_FLAGS = {
  deployedModel: {
@@ -72,6 +72,7 @@ export default defineCommand({
    } else if (format === "text") {
      const cap = deployment?.capacity !== undefined ? ` (capacity=${deployment.capacity})` : "";
      emitBare(`Scaled ${deployedModel}${cap}.`);
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -4,7 +4,7 @@ import {
  updateDeployment,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const UPDATE_FLAGS = {
  deployedModel: {
@@ -70,6 +70,7 @@ export default defineCommand({
      if (deployment?.tpm_limit !== undefined) parts.push(`tpm_limit=${deployment.tpm_limit}`);
      const summary = parts.length ? ` (${parts.join(", ")})` : "";
      emitBare(`Updated ${deployedModel}${summary}.`);
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -23,7 +23,7 @@ export default defineCommand({
    "--file photo.jpg --model qwen3-vl-plus",
    "--file video.mp4 --model wan2.1-t2v-plus",
    "--file audio.wav --model qwen3-asr-flash",
-    "--file cat.png --model qwen-image-2.0",
+    "--file cat.png --model qwen-image-3.0",
  ],
  async run(ctx) {
    const { settings, flags } = ctx;
@@ -1,5 +1,5 @@
 import { defineCommand, detectOutputFormat, cancelFineTune, type FlagsDef } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const CANCEL_FLAGS = {
  jobId: {
@@ -38,6 +38,7 @@ export default defineCommand({
    } else if (format === "text") {
      const status = job?.status ? ` (status=${job.status})` : "";
      emitBare(`Cancelled ${jobId}${status}.`);
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -4,7 +4,7 @@ import {
  listCheckpoints,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare, formatTable } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime";

 const CHECKPOINTS_FLAGS = {
  jobId: {
@@ -47,7 +47,7 @@ export default defineCommand({
    }));

    if (format === "json") {
-      emitResult({ items, total }, format);
+      emitResult({ items, total, request_id: response.request_id }, format);
      return;
    }

@@ -60,5 +60,6 @@ export default defineCommand({
    const rows = items.map((i) => [i.checkpoint, i.step, i.status]);
    for (const line of formatTable(headers, rows)) emitBare(line);
    emitBare(`\nTotal: ${total}`);
+    emitRequestId(response.request_id, settings.quiet);
  },
 });
@@ -27,7 +27,7 @@ import {
 } from "bailian-cli-core";
 import { existsSync, statSync } from "fs";
 import { basename } from "path";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 /**
 * A `--datasets` / `--validations` token is treated as a local file to upload
@@ -631,6 +631,7 @@ async function runCreate<F extends FlagsDef>(
    if (job?.job_id) {
      emitBare(`Created fine-tune job: ${job.job_id}`);
      if (job.status) emitBare(`Status: ${job.status}`);
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -1,5 +1,5 @@
 import { defineCommand, detectOutputFormat, deleteFineTune, type FlagsDef } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const DELETE_FLAGS = {
  jobId: {
@@ -36,6 +36,7 @@ export default defineCommand({
      emitBare(jobId);
    } else if (format === "text") {
      emitBare(`Deleted ${jobId}.`);
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -4,7 +4,7 @@ import {
  exportCheckpoint,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const EXPORT_FLAGS = {
  jobId: {
@@ -69,6 +69,7 @@ export default defineCommand({
      emitBare(
        `Next: ${identity.binName} deploy text create --model ${exported} --name <display-name>`,
      );
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -1,5 +1,5 @@
 import { defineCommand, detectOutputFormat, getFineTune, type FlagsDef } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const GET_FLAGS = {
  jobId: {
@@ -56,7 +56,7 @@ export default defineCommand({
    };

    if (format === "json") {
-      emitResult(item, format);
+      emitResult({ ...item, request_id: response.request_id }, format);
      return;
    }

@@ -76,5 +76,6 @@ export default defineCommand({
    if (item.model_name) emitBare(`model_name:       ${item.model_name}`);
    if (item.created_at) emitBare(`created_at:       ${item.created_at}`);
    if (item.updated_at) emitBare(`updated_at:       ${item.updated_at}`);
+    emitRequestId(response.request_id, settings.quiet);
  },
 });
@@ -1,5 +1,5 @@
 import { defineCommand, detectOutputFormat, listFineTunes, type FlagsDef } from "bailian-cli-core";
-import { emitResult, emitBare, formatTable } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId, formatTable } from "bailian-cli-runtime";

 const LIST_FLAGS = {
  page: { type: "number", valueHint: "<n>", description: "Page number (default: 1)" },
@@ -48,7 +48,7 @@ export default defineCommand({
    }));

    if (format === "json") {
-      emitResult({ items, total }, format);
+      emitResult({ items, total, request_id: response.request_id }, format);
      return;
    }

@@ -78,5 +78,6 @@ export default defineCommand({
    emitBare(
      `Tip: OUTPUT_MODEL is the input for \`${identity.binName} deploy text create --model\``,
    );
+    emitRequestId(response.request_id, settings.quiet);
  },
 });
@@ -6,7 +6,7 @@ import {
  type FineTuneLogEntry,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 /**
 * Render a single log entry as a single line (mirrors the flatten logic used
@@ -187,6 +187,7 @@ export default defineCommand({
        emitBare(renderEntry(entry));
      }
      if (payload?.total !== undefined) emitBare(`\nTotal: ${payload.total}`);
+      emitRequestId(response.request_id, settings.quiet);
    } else {
      emitResult(response, format);
    }
@@ -6,7 +6,7 @@ import {
  ExitCode,
  type FlagsDef,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "bailian-cli-runtime";
+import { emitResult, emitBare, emitRequestId } from "bailian-cli-runtime";

 const DEFAULT_INTERVAL_SEC = 10;
 const MIN_INTERVAL_SEC = 1;
@@ -135,9 +135,13 @@ export default defineCommand({
      } else if (format === "text") {
        emitBare(`${nowStamp()}  ${jobId}  ${status || "UNKNOWN"}`);
        if (status === "SUCCEEDED") emitBare(`✓ ${jobId}  ${status}`);
+        emitRequestId(response.request_id, settings.quiet);
      } else {
        // json: a compact, purpose-built status probe.
-        emitResult({ job_id: jobId, status: status || "UNKNOWN", terminal }, format);
+        emitResult(
+          { job_id: jobId, status: status || "UNKNOWN", terminal, request_id: response.request_id },
+          format,
+        );
      }

      if (terminal && status !== "SUCCEEDED") {
@@ -175,6 +179,7 @@ export default defineCommand({
            emitResult(response, format);
          } else if (status === "SUCCEEDED") {
            emitBare(`\n✓ ${jobId}  ${status}  (elapsed ${formatElapsed(elapsed)})`);
+            emitRequestId(response.request_id, settings.quiet);
          }
          if (status !== "SUCCEEDED") {
            throw new BailianError(
@@ -47,7 +47,7 @@ const EDIT_FLAGS = {
  model: {
    type: "string",
    valueHint: "<model>",
-    description: "Model ID (default: qwen-image-2.0)",
+    description: "Model ID (default: qwen-image-3.0)",
  },
  size: {
    type: "string",
@@ -123,7 +123,7 @@ export default defineCommand({
    }
    const prompt = flags.prompt;

-    const model = flags.model || settings.defaultImageModel || "qwen-image-2.0";
+    const model = flags.model || settings.defaultImageModel || "qwen-image-3.0";
    const route = resolveImageEditApi(model);

    // Auto-upload local files (resolve all images in parallel)
@@ -35,7 +35,7 @@ const GENERATE_FLAGS = {
  model: {
    type: "string",
    valueHint: "<model>",
-    description: "Model ID (default: qwen-image-2.0)",
+    description: "Model ID (default: qwen-image-3.0)",
  },
  size: {
    type: "string",
@@ -105,7 +105,7 @@ export default defineCommand({
    const { settings, flags } = ctx;
    const prompt = flags.prompt;

-    const model = flags.model || settings.defaultImageModel || "qwen-image-2.0";
+    const model = flags.model || settings.defaultImageModel || "qwen-image-3.0";
    const route = resolveImageGenerateApi(model);
    const defaultSize = "1:1";
    const sizeInput = flags.size || defaultSize;
@@ -0,0 +1,122 @@
+import { mkdirSync } from "node:fs";
+import { dirname, join } from "node:path";
+import {
+  type BackendRuntimeInput,
+  LocalFileStateBackend,
+  resolveProjectConfigFromObject,
+} from "@openagentpack/sdk";
+import { getConfigDir } from "bailian-cli-core";
+import {
+  assertProviderCredentials,
+  type CredentialHost,
+  injectProviderCredentials,
+  normalizeInterpolatedProviderBlocks,
+  prepareProviderEnv,
+  scrubCredentialEnv,
+} from "./credentials.ts";
+import { type HostContext, installSdkTransport } from "./transport.ts";
+
+/** Default agent identity `bl managed-agent run` materializes and reuses. */
+export const DEFAULT_INLINE_AGENT = "dsh-remote-runner";
+
+/** Default model for the materialized agent. */
+export const DEFAULT_INLINE_MODEL = "qwen3.8-max";
+
+/** Default role when the caller supplies no `--instructions`. */
+export const DEFAULT_INLINE_INSTRUCTIONS = "You are a helpful assistant. Complete the task.";
+
+/** Environment name declared in the inline config; one cloud env per agent. */
+const INLINE_ENVIRONMENT = "cloud";
+
+export interface InlineAgentOptions {
+  agentName: string;
+  instructions: string;
+  model: string;
+  /** Override the persisted state location (defaults under the bl config dir). */
+  statePath?: string;
+}
+
+/**
+ * Slugify an agent name into a filesystem- and project-id-safe token. The state
+ * for each distinct agent lives in its own directory so repeat runs reuse the
+ * same materialized remote agent.
+ */
+function slugify(agentName: string): string {
+  const slug = agentName
+    .toLowerCase()
+    .replace(/[^a-z0-9._-]+/g, "-")
+    .replace(/^-+|-+$/g, "");
+  return slug.length > 0 ? slug : "agent";
+}
+
+/** Where a materialized agent's state is persisted (not the user's cwd). */
+export function inlineStatePath(agentName: string): string {
+  return join(getConfigDir(), "managed-agent", slugify(agentName), "state.json");
+}
+
+/**
+ * The minimal in-memory project config that materializes into one cloud agent.
+ * `providers.bailian` carries empty `api_key`/`base_url` placeholders so
+ * {@link injectProviderCredentials} fills them from bl's auth chain (it only
+ * writes fields the block already declares).
+ */
+export function buildInlineConfig(opts: InlineAgentOptions): Record<string, unknown> {
+  return {
+    version: "1",
+    providers: {
+      bailian: { api_key: "", base_url: "" },
+    },
+    defaults: { provider: "bailian" },
+    environments: {
+      [INLINE_ENVIRONMENT]: {
+        description: "Bailian CLI cloud environment",
+        config: { type: "cloud", networking: { type: "unrestricted" } },
+      },
+    },
+    agents: {
+      [opts.agentName]: {
+        description: opts.agentName,
+        model: opts.model,
+        instructions: opts.instructions,
+        environment: INLINE_ENVIRONMENT,
+        provider: "bailian",
+      },
+    },
+  };
+}
+
+/**
+ * Build the `BackendRuntimeInput` shared by ensure (`syncAgentResourcesWith
+ * StateBackend`) and run (`readProjectRuntime` + `startSessionRun`). Mirrors the
+ * credential spine of {@link buildAgentRuntime} but sources config from an
+ * in-memory object instead of a file, so no `agents.yaml` or `apply` is required.
+ */
+export async function buildInlineBackendInput(
+  host: HostContext & CredentialHost,
+  opts: InlineAgentOptions,
+): Promise<BackendRuntimeInput> {
+  installSdkTransport(host);
+  prepareProviderEnv();
+
+  const rawConfig = buildInlineConfig(opts);
+  const { config, projectName } = await resolveProjectConfigFromObject(rawConfig, {
+    projectName: slugify(opts.agentName),
+  });
+
+  normalizeInterpolatedProviderBlocks(config.providers);
+  injectProviderCredentials(config.providers, host);
+  scrubCredentialEnv();
+  assertProviderCredentials(config.providers);
+
+  const statePath = opts.statePath ?? inlineStatePath(opts.agentName);
+  mkdirSync(dirname(statePath), { recursive: true });
+  const stateBackend = new LocalFileStateBackend({ statePath });
+
+  return {
+    projectName,
+    config,
+    stateBackend,
+    stateScope: { projectId: slugify(opts.agentName) },
+    providers: config.providers,
+  };
+}
@@ -25,7 +25,7 @@ const PROVIDER_BLOCKS: Record<string, string> = {
 };

 const SINGLE_MODEL: Record<string, string> = {
-  bailian: `    model: qwen3.7-max`,
+  bailian: `    model: qwen3.8-max`,
  claude: `    model: claude-sonnet-4-6`,
  qoder: `    model: ultimate`,
  ark: `    model: doubao-seed-2-1-pro-260628`,
@@ -39,7 +39,7 @@ function buildTemplate(options: { provider: string; agentName: string }): string

  const modelBlock =
    options.provider === "all"
-      ? `    model:\n      bailian: qwen3.7-max\n      claude: claude-sonnet-4-6\n      qoder: ultimate\n      ark: doubao-seed-2-1-pro-260628`
+      ? `    model:\n      bailian: qwen3.8-max\n      claude: claude-sonnet-4-6\n      qoder: ultimate\n      ark: doubao-seed-2-1-pro-260628`
      : SINGLE_MODEL[options.provider]!;

  const toolBlock =
@@ -0,0 +1,137 @@
+import {
+  BailianError,
+  defineCommand,
+  detectOutputFormat,
+  ExitCode,
+  type FlagsDef,
+} from "bailian-cli-core";
+import { emitResult } from "bailian-cli-runtime";
+import {
+  readProjectRuntime,
+  startSessionRun,
+  startSessionRunPolling,
+  syncAgentResourcesWithStateBackend,
+} from "@openagentpack/sdk";
+import { CREDENTIALS_NOTE } from "./_engine/config-loader.ts";
+import { withStdoutProtected } from "./_engine/console-capture.ts";
+import { withAgentErrors } from "./_engine/errors.ts";
+import {
+  buildInlineBackendInput,
+  DEFAULT_INLINE_AGENT,
+  DEFAULT_INLINE_INSTRUCTIONS,
+  DEFAULT_INLINE_MODEL,
+} from "./_engine/inline-runtime.ts";
+import { renderCollectedEvents, streamAndRenderEvents } from "./_engine/session-render.ts";
+
+const RUN_FLAGS = {
+  prompt: {
+    type: "string",
+    valueHint: "<text>",
+    description: "Task to run (required)",
+    required: true,
+  },
+  instructions: {
+    type: "string",
+    valueHint: "<text>",
+    description: "Role/system instructions for the remote agent (default: generic assistant)",
+  },
+  model: {
+    type: "string",
+    valueHint: "<id>",
+    description: `Model for the remote agent (default: ${DEFAULT_INLINE_MODEL})`,
+  },
+  agent: {
+    type: "string",
+    valueHint: "<name>",
+    description: `Agent identity to create/reuse (default: ${DEFAULT_INLINE_AGENT})`,
+  },
+  noStream: {
+    type: "switch",
+    description: "Use polling instead of SSE streaming",
+  },
+} satisfies FlagsDef;
+
+export default defineCommand({
+  description: "Provision (if needed) a cloud agent and run a task in one step",
+  auth: "apiKey",
+  usageArgs: "--prompt <text> [--instructions <text>] [--model <id>] [--agent <name>]",
+  flags: RUN_FLAGS,
+  exampleArgs: [
+    '--prompt "Summarize the latest AI news"',
+    '--prompt "Audit this dependency tree" --instructions "You are a security expert" --model qwen3.8-max',
+  ],
+  notes: [
+    ...CREDENTIALS_NOTE,
+    "Unlike `apply`, this creates/updates the cloud agent + environment on demand without --yes. The first run provisions cloud resources (may incur cost and take longer to start); later runs with the same --agent reuse them.",
+  ],
+  async run(ctx) {
+    const { settings, flags } = ctx;
+    const format = detectOutputFormat(settings.output);
+    const asJson = format === "json";
+
+    const agentName = flags.agent ?? DEFAULT_INLINE_AGENT;
+    const model = flags.model ?? DEFAULT_INLINE_MODEL;
+    const instructions = flags.instructions ?? DEFAULT_INLINE_INSTRUCTIONS;
+
+    if (settings.dryRun) {
+      emitResult(
+        {
+          would_run: {
+            prompt: flags.prompt,
+            agent: agentName,
+            model,
+            instructions,
+            mode: flags.noStream ? "polling" : "streaming",
+          },
+        },
+        format,
+      );
+      return;
+    }
+
+    await withAgentErrors(() =>
+      withStdoutProtected(async () => {
+        const input = await buildInlineBackendInput(ctx, { agentName, instructions, model });
+
+        // Ensure the remote agent + its cloud environment exist. Idempotent:
+        // a repeat run with the same agent name reuses the materialized state.
+        if (!asJson) process.stderr.write(`Ensuring cloud agent "${agentName}"…\n`);
+        const sync = await syncAgentResourcesWithStateBackend(input, agentName, {
+          policy: "force",
+          quiet: true,
+        });
+        if (sync.status !== "completed") {
+          const detail =
+            sync.error ??
+            sync.diagnostics.find((diag) => diag.severity === "error")?.message ??
+            `provisioning ended with status "${sync.status}"`;
+          throw new BailianError(
+            `Failed to provision cloud agent "${agentName}": ${detail}`,
+            ExitCode.GENERAL,
+          );
+        }
+
+        // Run the task inside a runtime bound to the just-materialized state.
+        await readProjectRuntime(input, async (runtime) => {
+          if (flags.noStream) {
+            const run = await startSessionRunPolling(runtime, flags.prompt, { agent: agentName });
+            if (!asJson) process.stderr.write(`Session created: ${run.session.id}\n`);
+            renderCollectedEvents(run, asJson, {
+              session_id: run.session.id,
+              provider: run.provider,
+              agent: run.agentName,
+            });
+          } else {
+            const run = await startSessionRun(runtime, flags.prompt, { agent: agentName });
+            if (!asJson) process.stderr.write(`Session created: ${run.session.id}\n`);
+            await streamAndRenderEvents(run.events, asJson, {
+              session_id: run.session.id,
+              provider: run.provider,
+              agent: run.agentName,
+            });
+          }
+        });
+      }),
+    );
+  },
+});
@@ -28,6 +28,16 @@ const ADD_FLAGS = {
    valueHint: "<id>",
    description: "Memory library ID (isolate memory space)",
  },
+  projectId: {
+    type: "string",
+    valueHint: "<id>",
+    description: "Memory extraction rule ID (defaults to the library's default rule)",
+  },
+  metaData: {
+    type: "string",
+    valueHint: "<json>",
+    description: 'Custom metadata JSON object: {"location":"Beijing"}',
+  },
 } satisfies FlagsDef;
 type AddFlags = ParsedFlags<typeof ADD_FLAGS>;

@@ -40,6 +50,7 @@ export default defineCommand({
    '--user-id user1 --content "The user likes Python programming"',
    '--user-id user1 --messages \'[{"role":"user","content":"I like traveling"}]\'',
    '--user-id user1 --content "Lives in Beijing" --profile-schema schema_xxx',
+    '--user-id user1 --content "Lives in Beijing" --meta-data \'{"source":"onboarding"}\'',
  ],
  validate: (f: AddFlags) =>
    !f.messages && !f.content ? "Provide --messages or --content." : undefined,
@@ -63,6 +74,15 @@ export default defineCommand({

    if (flags.profileSchema) body.profile_schema = flags.profileSchema;
    if (flags.memoryLibraryId) body.memory_library_id = flags.memoryLibraryId;
+    if (flags.projectId) body.project_id = flags.projectId;
+
+    if (flags.metaData) {
+      try {
+        body.meta_data = JSON.parse(flags.metaData);
+      } catch {
+        throw new UsageError("--meta-data must be valid JSON object");
+      }
+    }

    const format = detectOutputFormat(settings.output);

@@ -78,8 +98,14 @@ export default defineCommand({
    });

    if (settings.quiet || format === "text") {
-      const ids = response.memory_ids?.join(", ") || "none";
-      emitBare(`Memory added. IDs: ${ids}`);
+      const nodes = response.memory_nodes ?? [];
+      if (nodes.length === 0) {
+        emitBare("No memory fragments were extracted.");
+      } else {
+        for (const node of nodes) {
+          emitBare(`[${node.event ?? "ADD"}] ${node.memory_node_id} ${node.content}`);
+        }
+      }
    } else {
      emitResult(response, format);
    }
@@ -24,6 +24,11 @@ export default defineCommand({
    },
    page: { type: "number", valueHint: "<n>", description: "Page number (default: 1)" },
    memoryLibraryId: { type: "string", valueHint: "<id>", description: "Memory library ID" },
+    projectId: {
+      type: "string",
+      valueHint: "<id>",
+      description: "Memory extraction rule ID (defaults to the library's default rule)",
+    },
  },
  exampleArgs: ["--user-id user1", "--user-id user1 --page-size 20 --page 2"],
  async run(ctx) {
@@ -36,6 +41,7 @@ export default defineCommand({
    if (flags.pageSize !== undefined) params.set("page_size", String(flags.pageSize));
    if (flags.page !== undefined) params.set("page_num", String(flags.page));
    if (flags.memoryLibraryId) params.set("memory_library_id", flags.memoryLibraryId);
+    if (flags.projectId) params.set("project_id", flags.projectId);

    const path = `${memoryListPath()}?${params.toString()}`;

@@ -0,0 +1,44 @@
+import { defineCommand, profileSchemaItemPath, detectOutputFormat } from "bailian-cli-core";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "Delete a profile schema",
+  auth: "apiKey",
+  usageArgs: "--schema-id <id> [flags]",
+  flags: {
+    schemaId: {
+      type: "string",
+      valueHint: "<id>",
+      description: "Profile schema ID (required)",
+      required: true,
+    },
+    memoryLibraryId: { type: "string", valueHint: "<id>", description: "Memory library ID" },
+  },
+  exampleArgs: ["--schema-id schema_xxx"],
+  async run(ctx) {
+    const { settings, flags } = ctx;
+    const format = detectOutputFormat(settings.output);
+
+    const params = new URLSearchParams();
+    if (flags.memoryLibraryId) params.set("memory_library_id", flags.memoryLibraryId);
+    const query = params.toString();
+    const base = profileSchemaItemPath(flags.schemaId);
+    const path = query ? `${base}?${query}` : base;
+
+    if (settings.dryRun) {
+      emitResult({ endpoint: ctx.client.url(path), method: "DELETE" }, format);
+      return;
+    }
+
+    const response = await ctx.client.requestJson<{ request_id: string }>({
+      path,
+      method: "DELETE",
+    });
+
+    if (settings.quiet || format === "text") {
+      emitBare(`Profile schema ${flags.schemaId} deleted.`);
+    } else {
+      emitResult(response, format);
+    }
+  },
+});
@@ -0,0 +1,52 @@
+import {
+  defineCommand,
+  profileSchemaItemPath,
+  detectOutputFormat,
+  type ProfileSchemaGetResponse,
+} from "bailian-cli-core";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "Show a profile schema and its attribute IDs",
+  auth: "apiKey",
+  usageArgs: "--schema-id <id> [flags]",
+  flags: {
+    schemaId: {
+      type: "string",
+      valueHint: "<id>",
+      description: "Profile schema ID (required)",
+      required: true,
+    },
+    memoryLibraryId: { type: "string", valueHint: "<id>", description: "Memory library ID" },
+  },
+  exampleArgs: ["--schema-id schema_xxx"],
+  async run(ctx) {
+    const { settings, flags } = ctx;
+    const format = detectOutputFormat(settings.output);
+
+    const params = new URLSearchParams();
+    if (flags.memoryLibraryId) params.set("memory_library_id", flags.memoryLibraryId);
+    const query = params.toString();
+    const base = profileSchemaItemPath(flags.schemaId);
+    const path = query ? `${base}?${query}` : base;
+
+    if (settings.dryRun) {
+      emitResult({ endpoint: ctx.client.url(path), method: "GET" }, format);
+      return;
+    }
+
+    const response = await ctx.client.requestJson<ProfileSchemaGetResponse>({
+      path,
+      method: "GET",
+    });
+
+    if (settings.quiet || format === "text") {
+      emitBare(`${response.name}${response.description ? ` — ${response.description}` : ""}`);
+      for (const attribute of response.attributes ?? []) {
+        emitBare(`  [${attribute.attribute_id}] ${attribute.name}`);
+      }
+    } else {
+      emitResult(response, format);
+    }
+  },
+});
@@ -0,0 +1,55 @@
+import {
+  defineCommand,
+  profileSchemaPath,
+  detectOutputFormat,
+  type ProfileSchemaListResponse,
+} from "bailian-cli-core";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "List profile schemas",
+  auth: "apiKey",
+  usageArgs: "[flags]",
+  flags: {
+    memoryLibraryId: { type: "string", valueHint: "<id>", description: "Memory library ID" },
+    pageSize: { type: "number", valueHint: "<n>", description: "Results per page (default: 10)" },
+    page: { type: "number", valueHint: "<n>", description: "Page number (default: 1)" },
+  },
+  exampleArgs: ["", "--page-size 20 --page 2"],
+  async run(ctx) {
+    const { settings, flags } = ctx;
+    const format = detectOutputFormat(settings.output);
+
+    const params = new URLSearchParams();
+    if (flags.memoryLibraryId) params.set("memory_library_id", flags.memoryLibraryId);
+    if (flags.pageSize !== undefined) params.set("page_size", String(flags.pageSize));
+    if (flags.page !== undefined) params.set("page_num", String(flags.page));
+
+    const query = params.toString();
+    const path = query ? `${profileSchemaPath()}?${query}` : profileSchemaPath();
+
+    if (settings.dryRun) {
+      emitResult({ endpoint: ctx.client.url(path), method: "GET" }, format);
+      return;
+    }
+
+    const response = await ctx.client.requestJson<ProfileSchemaListResponse>({
+      path,
+      method: "GET",
+    });
+
+    if (settings.quiet || format === "text") {
+      const schemas = response.profile_schemas ?? [];
+      if (schemas.length === 0) {
+        emitBare("No profile schemas found.");
+      } else {
+        for (const schema of schemas) {
+          emitBare(`[${schema.profile_schema_id}] ${schema.name}`);
+        }
+        if (response.total !== undefined) emitBare(`\nTotal: ${response.total}`);
+      }
+    } else {
+      emitResult(response, format);
+    }
+  },
+});
@@ -0,0 +1,81 @@
+import {
+  defineCommand,
+  UsageError,
+  profileSchemaItemPath,
+  detectOutputFormat,
+  type ProfileSchemaUpdateRequest,
+} from "bailian-cli-core";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+import type { FlagsDef, ParsedFlags } from "bailian-cli-core";
+
+const UPDATE_FLAGS = {
+  schemaId: {
+    type: "string",
+    valueHint: "<id>",
+    description: "Profile schema ID (required)",
+    required: true,
+  },
+  name: { type: "string", valueHint: "<name>", description: "New schema name" },
+  description: { type: "string", valueHint: "<text>", description: "New schema description" },
+  attributeOps: {
+    type: "string",
+    valueHint: "<json>",
+    description:
+      'Attribute operations JSON array: [{"op":"add","name":"plan"},{"op":"delete","attribute_id":"attr_1"}]',
+  },
+  memoryLibraryId: { type: "string", valueHint: "<id>", description: "Memory library ID" },
+} satisfies FlagsDef;
+type UpdateFlags = ParsedFlags<typeof UPDATE_FLAGS>;
+
+export default defineCommand({
+  description: "Update a profile schema's name, description, or attributes",
+  auth: "apiKey",
+  usageArgs: "--schema-id <id> [--name <name>] [--attribute-ops <json>] [flags]",
+  flags: UPDATE_FLAGS,
+  notes: ["Attribute IDs for update/delete operations come from `memory profile detail`."],
+  exampleArgs: [
+    '--schema-id schema_xxx --name "user_basic_v2"',
+    '--schema-id schema_xxx --attribute-ops \'[{"op":"add","name":"plan","description":"subscription plan"}]\'',
+    '--schema-id schema_xxx --attribute-ops \'[{"op":"delete","attribute_id":"attr_1"}]\'',
+  ],
+  validate: (f: UpdateFlags) =>
+    !f.name && !f.description && !f.attributeOps
+      ? "Provide --name, --description, or --attribute-ops."
+      : undefined,
+  async run(ctx) {
+    const { settings, flags } = ctx;
+    const format = detectOutputFormat(settings.output);
+
+    const body: ProfileSchemaUpdateRequest = {};
+    if (flags.name) body.name = flags.name;
+    if (flags.description) body.description = flags.description;
+    if (flags.memoryLibraryId) body.memory_library_id = flags.memoryLibraryId;
+
+    if (flags.attributeOps) {
+      try {
+        body.attributes_operations = JSON.parse(flags.attributeOps);
+      } catch {
+        throw new UsageError("--attribute-ops must be valid JSON array");
+      }
+    }
+
+    const path = profileSchemaItemPath(flags.schemaId);
+
+    if (settings.dryRun) {
+      emitResult({ endpoint: ctx.client.url(path), method: "PATCH", request: body }, format);
+      return;
+    }
+
+    const response = await ctx.client.requestJson<{ request_id: string }>({
+      path,
+      method: "PATCH",
+      body,
+    });
+
+    if (settings.quiet || format === "text") {
+      emitBare(`Profile schema ${flags.schemaId} updated.`);
+    } else {
+      emitResult(response, format);
+    }
+  },
+});
@@ -24,6 +24,38 @@ const SEARCH_FLAGS = {
    description: "Number of results to return (default: 10)",
  },
  memoryLibraryId: { type: "string", valueHint: "<id>", description: "Memory library ID" },
+  projectIds: {
+    type: "array",
+    valueHint: "<id>",
+    description: "Memory extraction rule ID for hybrid retrieval (repeatable)",
+  },
+  minScore: {
+    type: "number",
+    valueHint: "<n>",
+    description: "Minimum similarity score, 0-1 (default: 0.3)",
+  },
+  enableRerank: {
+    type: "boolean",
+    valueHint: "<bool>",
+    description:
+      "Rerank results. Also selects the billing tier: false bills lite, true bills pro (~50x). (default: true)",
+  },
+  planVersion: {
+    type: "string",
+    valueHint: "<lite|pro>",
+    description:
+      "Documented billing tier. The service currently honors --enable-rerank instead, so prefer that flag",
+  },
+  enableJudge: {
+    type: "boolean",
+    valueHint: "<bool>",
+    description: "Enable the intent-discrimination callback (default: false)",
+  },
+  enableRewrite: {
+    type: "boolean",
+    valueHint: "<bool>",
+    description: "Enable query rewriting (default: false)",
+  },
 } satisfies FlagsDef;
 type SearchFlags = ParsedFlags<typeof SEARCH_FLAGS>;

@@ -35,6 +67,7 @@ export default defineCommand({
  exampleArgs: [
    '--user-id user1 --query "programming preferences"',
    '--user-id user1 --messages \'[{"role":"user","content":"recommend a book"}]\' --top-k 5',
+    '--user-id user1 --query "preferences" --enable-rerank false --min-score 0.5',
  ],
  validate: (f: SearchFlags) =>
    !f.query && !f.messages ? "Provide --query or --messages." : undefined,
@@ -61,6 +94,21 @@ export default defineCommand({

    if (flags.topK !== undefined) body.top_k = flags.topK;
    if (flags.memoryLibraryId) body.memory_library_id = flags.memoryLibraryId;
+    if (flags.projectIds && flags.projectIds.length > 0) body.project_ids = flags.projectIds;
+    if (flags.minScore !== undefined) body.min_score = flags.minScore;
+    if (flags.enableRerank !== undefined) body.enable_rerank = flags.enableRerank;
+    if (flags.enableJudge !== undefined) body.enable_judge = flags.enableJudge;
+    if (flags.enableRewrite !== undefined) body.enable_rewrite = flags.enableRewrite;
+
+    if (flags.planVersion) {
+      if (flags.planVersion !== "lite" && flags.planVersion !== "pro") {
+        throw new UsageError("--plan-version must be lite or pro");
+      }
+      body.plan_version = flags.planVersion;
+      // The service ignores plan_version on its own, so mirror the intent onto
+      // the flag it does honor unless the caller set that one explicitly.
+      if (flags.enableRerank === undefined) body.enable_rerank = flags.planVersion === "pro";
+    }

    const format = detectOutputFormat(settings.output);

@@ -1,5 +1,6 @@
 import {
  defineCommand,
+  UsageError,
  memoryNodePath,
  detectOutputFormat,
  type MemoryNodeUpdateRequest,
@@ -34,6 +35,16 @@ export default defineCommand({
      valueHint: "<id>",
      description: "Memory library ID (non-default library)",
    },
+    timestamp: {
+      type: "number",
+      valueHint: "<unix-seconds>",
+      description: "When the remembered event happened (default: now)",
+    },
+    metaData: {
+      type: "string",
+      valueHint: "<json>",
+      description: 'Custom metadata JSON object, merged incrementally: {"source":"manual"}',
+    },
  },
  exampleArgs: ['--node-id node_xxx --user-id user1 --content "updated memory content"'],
  async run(ctx) {
@@ -47,6 +58,15 @@ export default defineCommand({
      custom_content: content,
    };
    if (flags.memoryLibraryId) body.memory_library_id = flags.memoryLibraryId;
+    if (flags.timestamp !== undefined) body.timestamp = flags.timestamp;
+
+    if (flags.metaData) {
+      try {
+        body.meta_data = JSON.parse(flags.metaData);
+      } catch {
+        throw new UsageError("--meta-data must be valid JSON object");
+      }
+    }

    const format = detectOutputFormat(settings.output);

@@ -0,0 +1,123 @@
+import {
+  BailianError,
+  ExitCode,
+  defineCommand,
+  detectInstalledAgents,
+  fetchSkillsIndex,
+  getSkillRegistryBaseUrl,
+  installSkillWithFanout,
+  parseSkillNames,
+  readSkillLock,
+  runWithConcurrency,
+  writeSkillLock,
+} from "bailian-cli-core";
+import { emitBare, emitResult, formatTable } from "bailian-cli-runtime";
+
+interface AddOutcome {
+  name: string;
+  status: "installed" | "failed";
+  publishedAt?: string;
+  agents?: string[];
+  reason?: string;
+}
+
+/** Max number of skills downloading/installing at the same time. */
+const INSTALL_CONCURRENCY = 3;
+
+export default defineCommand({
+  description: "Install skills from the Bailian skill registry into local agents",
+  auth: "none",
+  usageArgs: "--all | --name <name,...>",
+  flags: {
+    all: {
+      type: "switch",
+      description: "Install all skills from the registry",
+    },
+    name: {
+      type: "string",
+      valueHint: "<name,...>",
+      description: "Comma-separated skill names to install",
+    },
+  },
+  validate(flags) {
+    if (flags.all && flags.name) return "Use either --all or --name, not both";
+    if (!flags.all && !flags.name)
+      return "Specify --all to install everything or --name <name,...> for specific skills";
+    return undefined;
+  },
+  exampleArgs: ["--all", "--name spark-video,bailian-model-recommend"],
+  async run(ctx) {
+    const format = ctx.settings.outputExplicit ? ctx.settings.output : "json";
+    const index = await fetchSkillsIndex();
+    const remoteNames = Object.keys(index.skills);
+    const parsed = ctx.flags.all ? "all" : parseSkillNames(ctx.flags.name, false);
+    const names = parsed === "all" ? remoteNames : parsed;
+
+    const lock = readSkillLock();
+    const agents = detectInstalledAgents();
+
+    // collect-then-throw: a single skill failure only affects itself; successful ones are written to disk and lock as usual.
+    // Skills install concurrently (bounded by INSTALL_CONCURRENCY) — each writes to a disjoint canonical dir, unique tmpDir, and distinct lock key.
+    const tasks = names.map((name) => async (): Promise<AddOutcome> => {
+      const entry = index.skills[name];
+      if (!entry) {
+        return { name, status: "failed", reason: "skill not found in registry" };
+      }
+      try {
+        const record = await installSkillWithFanout(
+          name,
+          entry,
+          agents,
+          lock.skills[name]?.links ?? [],
+        );
+        lock.skills[name] = record.lockEntry;
+        return {
+          name,
+          status: "installed",
+          publishedAt: entry.publishedAt,
+          agents: record.linkedAgents,
+        };
+      } catch (err) {
+        return {
+          name,
+          status: "failed",
+          reason: err instanceof Error ? err.message : String(err),
+        };
+      }
+    });
+    const results = await runWithConcurrency(tasks, INSTALL_CONCURRENCY);
+    writeSkillLock(lock);
+
+    if (format === "json") {
+      emitResult(
+        {
+          registry: getSkillRegistryBaseUrl(),
+          agents: agents.map((agent) => agent.id),
+          skills: results,
+        },
+        format,
+      );
+    } else if (results.length === 0) {
+      emitBare("Skill registry is empty; no skills to install.");
+    } else {
+      const rows = results.map((result) => [
+        result.name,
+        result.status,
+        result.publishedAt ? result.publishedAt.slice(0, 10) : "-",
+        result.status === "installed" ? result.agents?.join(", ") || "-" : (result.reason ?? "-"),
+      ]);
+      for (const line of formatTable(["NAME", "STATUS", "PUBLISHED", "AGENTS / REASON"], rows)) {
+        emitBare(line);
+      }
+    }
+
+    const failed = results.filter((result) => result.status === "failed");
+    if (failed.length > 0) {
+      throw new BailianError(
+        `${failed.length}/${results.length} skill(s) failed to install`,
+        ExitCode.GENERAL,
+        "Check the reason for failed skills in the output; network failures can be retried with bl skill add",
+      );
+    }
+  },
+});
@@ -0,0 +1,107 @@
+import {
+  BailianError,
+  ExitCode,
+  defineCommand,
+  detectInstalledAgents,
+  fetchSkillsIndex,
+  getSkillRegistryBaseUrl,
+  installSkillWithFanout,
+  readSkillLock,
+  runWithConcurrency,
+  writeSkillLock,
+} from "bailian-cli-core";
+import { emitBare, emitResult, formatTable } from "bailian-cli-runtime";
+
+interface InitOutcome {
+  name: string;
+  status: "installed" | "failed";
+  publishedAt?: string;
+  agents?: string[];
+  reason?: string;
+}
+
+/** Prefix used to identify first-party Bailian skills in the registry. */
+const BAILIAN_PREFIX = "bailian-";
+
+/** Max number of skills downloading/installing at the same time. */
+const INIT_CONCURRENCY = 3;
+
+export default defineCommand({
+  description: "Install all bailian-* skills (one-shot bootstrap for new environments)",
+  auth: "none",
+  usageArgs: "",
+  exampleArgs: [""],
+  notes: [
+    "Fetches the registry index and installs every skill whose name starts with bailian-",
+    "Equivalent to: bl skill add --all (filtered to bailian-* skills)",
+  ],
+  async run(ctx) {
+    const format = ctx.settings.outputExplicit ? ctx.settings.output : "json";
+    const index = await fetchSkillsIndex();
+
+    // Discover all bailian-* skills from the live registry index
+    const names = Object.keys(index.skills).filter((name) => name.startsWith(BAILIAN_PREFIX));
+
+    const lock = readSkillLock();
+    const agents = detectInstalledAgents();
+
+    const tasks = names.map((name) => async (): Promise<InitOutcome> => {
+      const entry = index.skills[name];
+      try {
+        const record = await installSkillWithFanout(
+          name,
+          entry,
+          agents,
+          lock.skills[name]?.links ?? [],
+        );
+        lock.skills[name] = record.lockEntry;
+        return {
+          name,
+          status: "installed",
+          publishedAt: entry.publishedAt,
+          agents: record.linkedAgents,
+        };
+      } catch (err) {
+        return {
+          name,
+          status: "failed",
+          reason: err instanceof Error ? err.message : String(err),
+        };
+      }
+    });
+    const results = await runWithConcurrency(tasks, INIT_CONCURRENCY);
+    writeSkillLock(lock);
+
+    if (format === "json") {
+      emitResult(
+        {
+          registry: getSkillRegistryBaseUrl(),
+          agents: agents.map((agent) => agent.id),
+          skills: results,
+        },
+        format,
+      );
+    } else if (results.length === 0) {
+      emitBare("No bailian-* skills found in the registry.");
+    } else {
+      const rows = results.map((result) => [
+        result.name,
+        result.status,
+        result.publishedAt ? result.publishedAt.slice(0, 10) : "-",
+        result.status === "installed" ? result.agents?.join(", ") || "-" : (result.reason ?? "-"),
+      ]);
+      for (const line of formatTable(["NAME", "STATUS", "PUBLISHED", "AGENTS / REASON"], rows)) {
+        emitBare(line);
+      }
+    }
+
+    const failed = results.filter((result) => result.status === "failed");
+    if (failed.length > 0) {
+      throw new BailianError(
+        `${failed.length}/${results.length} skill(s) failed to install`,
+        ExitCode.GENERAL,
+        "Check the reason for failed skills in the output; network failures can be retried with bl skill init",
+      );
+    }
+  },
+});
@@ -0,0 +1,57 @@
+import {
+  defineCommand,
+  computeSkillStatuses,
+  fetchSkillsIndex,
+  getSkillRegistryBaseUrl,
+  listSkillDirsOnDisk,
+  readSkillLock,
+} from "bailian-cli-core";
+import { emitBare, emitResult, formatTable } from "bailian-cli-runtime";
+
+const DESCRIPTION_MAX = 60;
+
+function truncate(text: string | undefined): string {
+  if (!text) return "-";
+  return text.length > DESCRIPTION_MAX ? `${text.slice(0, DESCRIPTION_MAX - 1)}…` : text;
+}
+
+export default defineCommand({
+  description: "List registry skills and diff against local installs",
+  auth: "none",
+  exampleArgs: ["", "--output json"],
+  notes: [
+    "STATUS: installed | outdated | not-installed | missing (lock has it, dir deleted) | untracked (dir exists, not managed)",
+  ],
+  async run(ctx) {
+    const format = ctx.settings.outputExplicit ? ctx.settings.output : "json";
+    // Three-way reconciliation: live remote index × skill-lock.json (installation facts) × disk
+    const index = await fetchSkillsIndex();
+    const lock = readSkillLock();
+    const rows = computeSkillStatuses(index, lock, listSkillDirsOnDisk());
+
+    if (format === "json") {
+      emitResult(
+        {
+          registry: getSkillRegistryBaseUrl(),
+          ...(index.updatedAt ? { updatedAt: index.updatedAt } : {}),
+          skills: rows,
+        },
+        format,
+      );
+      return;
+    }
+    if (rows.length === 0) {
+      emitBare("Skill registry is empty and no skills are installed locally.");
+      return;
+    }
+    const table = rows.map((row) => [
+      row.name,
+      row.status,
+      row.publishedAt ? row.publishedAt.slice(0, 19).replace("T", " ") : "-",
+      truncate(row.description),
+    ]);
+    for (const line of formatTable(["NAME", "STATUS", "UPDATEDAT", "DESCRIPTION"], table)) {
+      emitBare(line);
+    }
+  },
+});
@@ -0,0 +1,99 @@
+import {
+  BailianError,
+  ExitCode,
+  defineCommand,
+  listSkillDirsOnDisk,
+  parseSkillNames,
+  readSkillLock,
+  removeSkillDir,
+  unlinkSkillFromAgents,
+  writeSkillLock,
+} from "bailian-cli-core";
+import { emitBare, emitResult, formatTable } from "bailian-cli-runtime";
+
+interface RemoveOutcome {
+  name: string;
+  status: "removed" | "failed";
+  removedLinks?: number;
+  reason?: string;
+}
+
+export default defineCommand({
+  description: "Remove locally installed skills (registry is untouched)",
+  auth: "none",
+  usageArgs: "--name <all|name,...>",
+  flags: {
+    name: {
+      type: "string",
+      valueHint: "<all|name,...>",
+      description: "Skills to remove: all or comma-separated skill names",
+      required: true,
+    },
+  },
+  exampleArgs: ["--name spark-video", "--name all"],
+  async run(ctx) {
+    // Purely local operation: no remote access, works offline
+    const format = ctx.settings.outputExplicit ? ctx.settings.output : "json";
+    const requested = parseSkillNames(ctx.flags.name, false);
+    const lock = readSkillLock();
+    const names = requested === "all" ? Object.keys(lock.skills) : requested;
+
+    if (names.length === 0) {
+      emitResult({ skills: [] }, format);
+      if (format === "text") emitBare("No skills installed locally; nothing to remove.");
+      return;
+    }
+
+    const diskDirs = new Set(listSkillDirsOnDisk());
+    const results: RemoveOutcome[] = [];
+    for (const name of names) {
+      const locked = lock.skills[name];
+      if (!locked) {
+        results.push({
+          name,
+          status: "failed",
+          reason: diskDirs.has(name)
+            ? "directory not managed by bl skill (untracked); remove manually if needed"
+            : "not installed",
+        });
+        continue;
+      }
+      try {
+        // Reclaim agent fan-out first, then delete canonical, finally clear the lock entry
+        const removedLinks = unlinkSkillFromAgents(name, locked.links ?? []);
+        removeSkillDir(name);
+        delete lock.skills[name];
+        results.push({ name, status: "removed", removedLinks: removedLinks.length });
+      } catch (err) {
+        results.push({
+          name,
+          status: "failed",
+          reason: err instanceof Error ? err.message : String(err),
+        });
+      }
+    }
+    writeSkillLock(lock);
+
+    if (format === "json") {
+      emitResult({ skills: results }, format);
+    } else {
+      const rows = results.map((r) => [
+        r.name,
+        r.status,
+        r.status === "removed" ? `reclaimed ${r.removedLinks} agent link(s)` : (r.reason ?? "-"),
+      ]);
+      for (const line of formatTable(["NAME", "STATUS", "DETAIL"], rows)) {
+        emitBare(line);
+      }
+    }
+
+    const failed = results.filter((r) => r.status === "failed");
+    if (failed.length > 0) {
+      throw new BailianError(
+        `${failed.length}/${results.length} skill(s) failed to remove`,
+        ExitCode.GENERAL,
+        "Check the reason for failed skills in the output; use bl skill list to verify local install status",
+      );
+    }
+  },
+});
@@ -0,0 +1,158 @@
+import {
+  BailianError,
+  ExitCode,
+  defineCommand,
+  detectInstalledAgents,
+  fanOutSkillToAgents,
+  fetchSkillsIndex,
+  getSkillRegistryBaseUrl,
+  installSkillWithFanout,
+  listSkillDirsOnDisk,
+  parseSkillNames,
+  readSkillLock,
+  runWithConcurrency,
+  writeSkillLock,
+} from "bailian-cli-core";
+import { emitBare, emitResult, formatTable } from "bailian-cli-runtime";
+
+interface UpdateOutcome {
+  name: string;
+  status: "updated" | "up-to-date" | "skipped" | "failed";
+  publishedAt?: string;
+  reason?: string;
+}
+
+/** Max number of skills downloading/installing at the same time. */
+const UPDATE_CONCURRENCY = 3;
+
+export default defineCommand({
+  description: "Update installed skills to the latest registry versions",
+  auth: "none",
+  usageArgs: "[--all] [--name <name,...>]",
+  flags: {
+    all: {
+      type: "switch",
+      description: "Update all installed skills (default when neither --all nor --name is given)",
+    },
+    name: {
+      type: "string",
+      valueHint: "<name,...>",
+      description: "Comma-separated skill names to update (must be already installed)",
+    },
+  },
+  validate(flags) {
+    if (flags.all && flags.name) return "Use either --all or --name, not both";
+    return undefined;
+  },
+  exampleArgs: ["", "--all", "--name spark-video"],
+  async run(ctx) {
+    const format = ctx.settings.outputExplicit ? ctx.settings.output : "json";
+    const updateAll = ctx.flags.all || !ctx.flags.name;
+    const requested = updateAll ? "all" : parseSkillNames(ctx.flags.name, false);
+    const index = await fetchSkillsIndex();
+    const lock = readSkillLock();
+    const disk = new Set(listSkillDirsOnDisk());
+
+    const agents = detectInstalledAgents();
+    const results: UpdateOutcome[] = [];
+    const targets: string[] = [];
+    if (requested === "all") {
+      // Default: only process skills already installed in lock; reinstall only if version changed or local dir is missing
+      for (const [name, locked] of Object.entries(lock.skills)) {
+        const entry = index.skills[name];
+        if (!entry) {
+          results.push({
+            name,
+            status: "skipped",
+            reason: "delisted from remote; local copy retained",
+          });
+          continue;
+        }
+        if (entry.contentHash === locked.contentHash && disk.has(name)) {
+          // Self-healing: content unchanged, but still fill fan-out links for agents
+          // detected since the last install (and refresh recorded copies); the merged
+          // ledger keeps paths of unvisited agents reclaimable by bl skill remove
+          const fanout = fanOutSkillToAgents(name, agents, locked.links ?? []);
+          lock.skills[name] = { ...locked, links: fanout.links };
+          results.push({ name, status: "up-to-date", publishedAt: locked.publishedAt });
+          continue;
+        }
+        targets.push(name);
+      }
+    } else {
+      // Explicit names: only update skills that are already installed; reject uninstalled ones
+      for (const name of requested) {
+        if (!lock.skills[name]) {
+          results.push({
+            name,
+            status: "failed",
+            reason: "not installed; run bl skill add --name " + name + " first",
+          });
+          continue;
+        }
+        targets.push(name);
+      }
+    }
+
+    const tasks = targets.map((name) => async (): Promise<UpdateOutcome> => {
+      const entry = index.skills[name];
+      if (!entry) {
+        return { name, status: "failed", reason: "skill not found in registry" };
+      }
+      try {
+        const record = await installSkillWithFanout(
+          name,
+          entry,
+          agents,
+          lock.skills[name]?.links ?? [],
+        );
+        lock.skills[name] = record.lockEntry;
+        return { name, status: "updated", publishedAt: entry.publishedAt };
+      } catch (err) {
+        return {
+          name,
+          status: "failed",
+          reason: err instanceof Error ? err.message : String(err),
+        };
+      }
+    });
+    const updateResults = await runWithConcurrency(tasks, UPDATE_CONCURRENCY);
+    results.push(...updateResults);
+    writeSkillLock(lock);
+
+    if (format === "json") {
+      emitResult({ registry: getSkillRegistryBaseUrl(), skills: results }, format);
+    } else if (results.length === 0) {
+      emitBare("No skills installed locally; run bl skill add first.");
+    } else {
+      const rows = results.map((result) => [
+        result.name,
+        result.status,
+        result.publishedAt ? result.publishedAt.slice(0, 10) : "-",
+      ]);
+      for (const line of formatTable(["NAME", "STATUS", "PUBLISHED"], rows)) {
+        emitBare(line);
+      }
+
+      // Footnotes for skipped / failed entries
+      const annotated = results.filter(
+        (result) => (result.status === "skipped" || result.status === "failed") && result.reason,
+      );
+      if (annotated.length > 0) {
+        emitBare("");
+        for (const result of annotated) {
+          emitBare(`  ${result.name}: ${result.reason}`);
+        }
+      }
+    }
+
+    const failed = results.filter((result) => result.status === "failed");
+    if (failed.length > 0) {
+      throw new BailianError(
+        `${failed.length} skill(s) failed to update`,
+        ExitCode.GENERAL,
+        "Check the reason for failed skills in the output; network failures can be retried with bl skill update",
+      );
+    }
+  },
+});
@@ -9,7 +9,6 @@ import {
  type DashScopeASRRequest,
  type DashScopeASRTaskResult,
  type DashScopeAsyncResponse,
-  trackingHeaders,
  stripUndefined,
  taskPath,
  speechRecognizePath,
@@ -201,9 +200,7 @@ async function handleAsyncMode(
    }

    // Fetch transcription JSON
-    const transRes = await fetch(subResult.transcription_url, {
-      headers: trackingHeaders(),
-    });
+    const transRes = await fetch(subResult.transcription_url);
    if (!transRes.ok) {
      throw new BailianError(
        `Failed to download transcription: HTTP ${transRes.status}`,
@@ -15,7 +15,7 @@ import { ansi, emitResult, emitBare } from "bailian-cli-runtime";
 import { readFileSync } from "fs";

 const CHAT_FLAGS = {
-  model: { type: "string", valueHint: "<model>", description: "Model ID (default: qwen3.7-max)" },
+  model: { type: "string", valueHint: "<model>", description: "Model ID (default: qwen3.8-max)" },
  message: {
    type: "array",
    valueHint: "<text>",
@@ -123,7 +123,7 @@ export default defineCommand({
    const { settings, flags } = ctx;
    const { system, messages } = parseMessages(flags);

-    const model = flags.model || settings.defaultTextModel || "qwen3.7-max";
+    const model = flags.model || settings.defaultTextModel || "qwen3.8-max";
    const shouldStream = flags.stream || process.stdout.isTTY;
    const format = detectOutputFormat(settings.output);

@@ -1,22 +1,31 @@
 import { execSync } from "child_process";
 import { writeFileSync } from "fs";
 import { join } from "path";
-import { defineCommand, getConfigDir } from "bailian-cli-core";
-import { ansi, fetchLatestVersion, type AnsiStyles } from "bailian-cli-runtime";
+import {
+  BailianError,
+  DEFAULT_INSTALL_PS1_URL,
+  DEFAULT_INSTALL_SCRIPT_URL,
+  defineCommand,
+  getConfigDir,
+  getUpdateInstallMethod,
+  type InstallMethod,
+} from "bailian-cli-core";
+import {
+  ansi,
+  fetchLatestVersion,
+  fetchBinaryChannelVersion,
+  isValidUpdateTargetVersion,
+  normalizeBinaryVersion,
+  performBinaryUpdate,
+  type AnsiStyles,
+} from "bailian-cli-runtime";

 const SKILL_SOURCE = "modelstudioai/cli";
 const SKILL_INSTALL_CMD = `npx skills add ${SKILL_SOURCE} --all -g -y`;

-/** Build the install command for the given npm package. */
-function detectInstallCommand(npmPackage: string): { cmd: string; label: string } {
-  return { cmd: `npm install -g ${npmPackage}@latest`, label: "npm" };
-}
-
 function updateAgentSkill(color: AnsiStyles): void {
  process.stderr.write("\nUpdating agent skill...\n");
  try {
-    // Reinstall (not `skills update`) into ~/.agents/skills/ and sync to all agent apps.
-    // `--all` on `skills add` means --skill '*' --agent '*' -y (Cursor, Claude Code, etc.).
    execSync(SKILL_INSTALL_CMD, { stdio: "inherit" });
    process.stderr.write(`${color.green("\u2713 Agent skill updated.")}\n`);
  } catch {
@@ -26,56 +35,140 @@ function updateAgentSkill(color: AnsiStyles): void {
  }
 }

+function writeUpdateState(version: string): void {
+  try {
+    const stateFile = join(getConfigDir(), "update-state.json");
+    writeFileSync(stateFile, JSON.stringify({ lastChecked: Date.now(), latestVersion: version }));
+  } catch {
+    /* ignore */
+  }
+}
+
+async function resolveLatest(method: InstallMethod, npmPackage: string): Promise<string | null> {
+  if (method === "binary") {
+    return (
+      (await fetchBinaryChannelVersion("latest", 5000)) ??
+      (await fetchLatestVersion(5000, npmPackage))
+    );
+  }
+  return fetchLatestVersion(5000, npmPackage);
+}
+
+function binaryReinstallHint(): string {
+  if (process.platform === "win32") {
+    return `  irm ${DEFAULT_INSTALL_PS1_URL} | iex\n`;
+  }
+  return `  curl -fsSL ${DEFAULT_INSTALL_SCRIPT_URL} | bash\n`;
+}
+
 export default defineCommand({
-  description: "Update the CLI to the latest version",
+  description: "Update the CLI to the latest or a specified version",
  auth: "none",
-  exampleArgs: [""],
+  usageArgs: "[--to <version>]",
+  flags: {
+    to: {
+      type: "string",
+      valueHint: "<version>",
+      description: "Install this exact version instead of the latest",
+    },
+  },
+  exampleArgs: ["", "--to 0.1.14"],
+  validate(flags) {
+    if (flags.to === undefined) return undefined;
+    if (!flags.to.trim()) return "--to requires a non-empty version";
+    if (!isValidUpdateTargetVersion(flags.to)) {
+      return `--to must be a semver version (e.g. 1.13.0, v1.13.0, 0.0.0-beta-<sha>-<YYYYMMDDHHMM>), got: ${flags.to.trim()}`;
+    }
+    return undefined;
+  },
  async run(ctx) {
    const { identity } = ctx;
    const npmPackage = identity.npmPackage;
    const binName = identity.binName;
    const currentVersion = identity.version;
    const color = ansi(process.stderr);
+    const method = getUpdateInstallMethod(identity);
+    const requestedTo = ctx.flags.to?.trim();
+    const pinnedVersion = requestedTo ? normalizeBinaryVersion(requestedTo) : undefined;

    process.stderr.write(`Current version: ${color.yellow(currentVersion)}\n`);
+    process.stderr.write(`Install method: ${color.dim(method)}\n`);
+    if (pinnedVersion) {
+      process.stderr.write(`Target version: ${color.green(pinnedVersion)}\n`);
+    } else {
+      process.stderr.write("Checking for updates...\n");
+    }

-    // Check latest version first
-    process.stderr.write("Checking for updates...\n");
-    const latest = await fetchLatestVersion(5000, npmPackage);
-
-    if (latest && latest === currentVersion) {
-      process.stderr.write(`${color.green(`\u2713 Already up to date (${currentVersion}).`)}\n`);
-      updateAgentSkill(color);
+    if (method === "brew" || method === "winget") {
+      const cmd =
+        method === "brew" ? "brew upgrade bailian-cli" : "winget upgrade Aliyun.BailianCLI";
+      process.stderr.write(
+        `${color.yellow(`This CLI was installed via ${method}. Update with:`)}\n  ${cmd}\n`,
+      );
+      if (pinnedVersion) {
+        process.stderr.write(
+          `${color.dim(`Note: --to is not supported for ${method} installs.`)}\n`,
+        );
+      }
      return;
    }

-    if (latest) {
-      process.stderr.write(`Latest version: ${color.green(latest)}\n\n`);
+    const targetVersion = pinnedVersion ?? (await resolveLatest(method, npmPackage));
+
+    if (!targetVersion) {
+      process.stderr.write(`${color.yellow("Could not determine the latest version.")}\n`);
+      return;
    }

-    const { cmd, label } = detectInstallCommand(npmPackage);
-    process.stderr.write(`Updating ${npmPackage} via ${label}...\n\n`);
+    if (targetVersion === currentVersion) {
+      const message = pinnedVersion
+        ? `\u2713 Already at ${currentVersion}.`
+        : `\u2713 Already up to date (${currentVersion}).`;
+      process.stderr.write(`${color.green(message)}\n`);
+      if (method === "npm") updateAgentSkill(color);
+      return;
+    }
+
+    if (!pinnedVersion) {
+      process.stderr.write(`Latest version: ${color.green(targetVersion)}\n\n`);
+    } else {
+      process.stderr.write("\n");
+    }
+
+    if (method === "binary") {
+      process.stderr.write(`Updating via binary channel...\n\n`);
+      try {
+        const newVer = await performBinaryUpdate(targetVersion);
+        process.stderr.write(
+          `\n${color.green(`\u2713 Update complete: ${currentVersion} \u2192 ${newVer}`)}\n`,
+        );
+        writeUpdateState(newVer);
+      } catch (error) {
+        const message = error instanceof Error ? error.message : String(error);
+        const reinstall =
+          error instanceof BailianError && error.hint
+            ? error.hint.replace(/^Re-run:\s*/i, "")
+            : binaryReinstallHint().trim();
+        process.stderr.write(`\nAutomatic binary update failed: ${message}\n`);
+        process.stderr.write("Re-run the install script:\n");
+        process.stderr.write(`  ${reinstall}\n\n`);
+      }
+      return;
+    }
+
+    const npmSpec = pinnedVersion ? `${npmPackage}@${pinnedVersion}` : `${npmPackage}@latest`;
+    const cmd = `npm install -g ${npmSpec}`;
+    process.stderr.write(`Updating ${npmPackage} via npm...\n\n`);

    try {
      execSync(cmd, { stdio: "inherit" });
-      // Verify the installed version after update
      try {
        const rawVer = execSync(`${binName} --version 2>/dev/null`, { encoding: "utf-8" }).trim();
-        // `<bin> --version` outputs "<bin> X.Y.Z" — extract just the version number
        const newVer = rawVer.replace(new RegExp(`^${binName}\\s+`), "");
        process.stderr.write(
          `\n${color.green(`\u2713 Update complete: ${currentVersion} \u2192 ${newVer}`)}\n`,
        );
-        // Update the cached state so the post-run notification doesn't fire
-        try {
-          const stateFile = join(getConfigDir(), "update-state.json");
-          writeFileSync(
-            stateFile,
-            JSON.stringify({ lastChecked: Date.now(), latestVersion: newVer }),
-          );
-        } catch {
-          /* ignore */
-        }
+        writeUpdateState(newVer);
      } catch {
        process.stderr.write(`\n${color.green("\u2713 Update complete.")}\n`);
      }
@@ -33,6 +33,10 @@ export { default as memoryUpdate } from "./commands/memory/update.ts";
 export { default as memoryDelete } from "./commands/memory/delete.ts";
 export { default as memoryProfileCreate } from "./commands/memory/profile-create.ts";
 export { default as memoryProfileGet } from "./commands/memory/profile-get.ts";
+export { default as memoryProfileList } from "./commands/memory/profile-list.ts";
+export { default as memoryProfileDetail } from "./commands/memory/profile-detail.ts";
+export { default as memoryProfileUpdate } from "./commands/memory/profile-update.ts";
+export { default as memoryProfileDelete } from "./commands/memory/profile-delete.ts";
 export { default as knowledgeRetrieve } from "./commands/knowledge/retrieve.ts";
 export { default as knowledgeSearch } from "./commands/knowledge/search.ts";
 export { default as knowledgeChat } from "./commands/knowledge/chat.ts";
@@ -95,6 +99,7 @@ export { default as managedAgentInit } from "./commands/managed-agent/init.ts";
 export { default as managedAgentValidate } from "./commands/managed-agent/validate.ts";
 export { default as managedAgentPlan } from "./commands/managed-agent/plan.ts";
 export { default as managedAgentApply } from "./commands/managed-agent/apply.ts";
+export { default as managedAgentRun } from "./commands/managed-agent/run.ts";
 export { default as managedAgentDestroy } from "./commands/managed-agent/destroy.ts";
 export { default as managedAgentStateList } from "./commands/managed-agent/state-list.ts";
 export { default as managedAgentStateShow } from "./commands/managed-agent/state-show.ts";
@@ -113,3 +118,8 @@ export { default as pluginInstall } from "./commands/plugin/install.ts";
 export { default as pluginLink } from "./commands/plugin/link.ts";
 export { default as pluginList } from "./commands/plugin/list.ts";
 export { default as pluginRemove } from "./commands/plugin/remove.ts";
+export { default as skillAdd } from "./commands/skill/add.ts";
+export { default as skillUpdate } from "./commands/skill/update.ts";
+export { default as skillRemove } from "./commands/skill/remove.ts";
+export { default as skillList } from "./commands/skill/list.ts";
+export { default as skillInit } from "./commands/skill/init.ts";
@@ -89,13 +89,13 @@ test("GET /api/config 返回全部 profile、明文密钥与持久化激活项",
    expect(res.json.enums.console_site).toEqual(["domestic", "international"]);
    expect(res.json.booleanKeys).toContain("telemetry");
    // Default field hints are surfaced as prefilled values in the UI.
-    expect(res.json.fieldDefaults.default_image_model).toBe("qwen-image-2.0");
-    expect(res.json.fieldDefaults.default_text_model).toBe("qwen3.7-max");
+    expect(res.json.fieldDefaults.default_image_model).toBe("qwen-image-3.0");
+    expect(res.json.fieldDefaults.default_text_model).toBe("qwen3.8-max");
    expect(res.json.fieldDefaults.output_dir).toContain("bailian-output");
    expect(res.json.fieldDefaults.timeout).toBe("300");
    expect(res.json.fieldDefaults.base_url).toBe("https://dashscope.aliyuncs.com");
    // Per-category model catalog (click-to-fill suggestions) is exposed too.
-    expect(res.json.modelCatalog.default_image_model[0]).toMatchObject({ id: "qwen-image-2.0" });
+    expect(res.json.modelCatalog.default_image_model[0]).toMatchObject({ id: "qwen-image-3.0" });
    expect(res.json.modelCatalog.default_video_model.map((m: { id: string }) => m.id)).toContain(
      "happyhorse-1.1-i2v",
    );
@@ -217,7 +217,7 @@ describe("e2e: auth", () => {
        authorization: "Bearer sk-e2e-placeholder",
        sourceConfig: expect.any(String),
        body: {
-          model: "qwen3.7-max",
+          model: "qwen3.8-max",
          stream: false,
        },
      });
@@ -266,7 +266,7 @@ describe("e2e: auth", () => {
      expect(config["token-plan"]).toMatchObject({
        api_key: "sk-sp-e2e-placeholder",
        base_url: validationServer.baseUrl,
-        default_text_model: "qwen3.8-max-preview",
+        default_text_model: "qwen3.8-max",
        default_video_model: "happyhorse-1.1-t2v",
        default_image_to_video_model: "happyhorse-1.1-i2v",
        default_reference_to_video_model: "happyhorse-1.1-r2v",
@@ -314,7 +314,7 @@ describe("e2e: auth", () => {
        authorization: "Bearer sk-sp-e2e-placeholder",
        sourceConfig: expect.any(String),
        body: {
-          model: "qwen3.7-max",
+          model: "qwen3.8-max",
          stream: false,
        },
      });
@@ -328,7 +328,7 @@ describe("e2e: auth", () => {
      expect(config["token-plan"]).toMatchObject({
        api_key: "sk-sp-e2e-placeholder",
        base_url: "https://token-plan.cn-beijing.maas.aliyuncs.com",
-        default_text_model: "qwen3.8-max-preview",
+        default_text_model: "qwen3.8-max",
        default_video_model: "happyhorse-1.1-t2v",
        default_image_to_video_model: "happyhorse-1.1-i2v",
        default_reference_to_video_model: "happyhorse-1.1-r2v",
@@ -21,7 +21,7 @@ agents:
  assistant:
    description: "E2E multi-provider fixture"
    model:
-      bailian: qwen3.7-max
+      bailian: qwen3.8-max
      claude: claude-sonnet-4-6
    instructions: |
      You are a helpful assistant.
@@ -194,13 +194,13 @@ describe.skipIf(!isBailianE2EMediaEnabled() || !isDashScopeE2EReady())("e2e: ima
    expect(stderr).toMatch(/--prompt|Usage:/i);
  });

-  test("【qwen-image-2.0】图片编辑", async () => {
+  test("【qwen-image-3.0】图片编辑", async () => {
    const outDir = makeE2eOutputDir(e2eLabelFromMetaUrl(import.meta.url));
    const gen = await runCommandE2e(IMAGE_ROUTES, [
      "image",
      "generate",
      "--model",
-      "qwen-image-2.0",
+      "qwen-image-3.0",
      "--prompt",
      "一只简笔画小猫，白底",
      "--out-dir",
@@ -220,7 +220,7 @@ describe.skipIf(!isBailianE2EMediaEnabled() || !isDashScopeE2EReady())("e2e: ima
      "image",
      "edit",
      "--model",
-      "qwen-image-2.0",
+      "qwen-image-3.0",
      "--image",
      imagePath!,
      "--prompt",
@@ -202,19 +202,19 @@ describe.skipIf(!isBailianE2EMediaEnabled() || !isDashScopeE2EReady())(
        "image",
        "generate",
        "--model",
-        "qwen-image-2.0",
+        "qwen-image-3.0",
      ]);
      expect(exitCode).toBe(2);
      expect(stderr).toMatch(/--prompt|Usage:/i);
    });

-    test("【qwen-image-2.0】图片生成", async () => {
+    test("【qwen-image-3.0】图片生成", async () => {
      const outDir = makeE2eOutputDir(e2eLabelFromMetaUrl(import.meta.url));
      const { stdout, stderr, exitCode } = await runCommandE2e(IMAGE_ROUTES, [
        "image",
        "generate",
        "--model",
-        "qwen-image-2.0",
+        "qwen-image-3.0",
        "--prompt",
        "一只简笔画小猫，白底",
        "--out-dir",
@@ -0,0 +1,148 @@
+import { existsSync, mkdtempSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+import { describe, expect, test } from "vite-plus/test";
+import { isBailianE2EEnabled, parseStdoutJson, runCommandE2e } from "./helpers.ts";
+import { SKILL_ROUTES } from "./topic-routes.ts";
+
+/** Canonical always-published skill; also the backbone of advisor wiki sync */
+const WIKI_SKILL = "bailian-docs-llm-wiki";
+
+/** Redirect ~/.bailian into a throwaway dir so lock/skill writes never touch the real user config */
+function makeTempConfigDir(): string {
+  return mkdtempSync(join(tmpdir(), "bl-skill-e2e-"));
+}
+
+describe("e2e: skill", () => {
+  test("skill add --help exits successfully", async () => {
+    const { stderr, exitCode } = await runCommandE2e(SKILL_ROUTES, ["skill", "add", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--all/);
+    expect(stderr).toMatch(/--name/);
+  });
+
+  test("skill update --help exits successfully", async () => {
+    const { stderr, exitCode } = await runCommandE2e(SKILL_ROUTES, ["skill", "update", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--all/);
+    expect(stderr).toMatch(/--name/);
+  });
+
+  test("skill remove --help exits successfully", async () => {
+    const { stderr, exitCode } = await runCommandE2e(SKILL_ROUTES, ["skill", "remove", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--name/);
+  });
+
+  test("skill list --help exits successfully", async () => {
+    const { stderr, exitCode } = await runCommandE2e(SKILL_ROUTES, ["skill", "list", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/list|registry/i);
+  });
+
+  test("skill init --help exits successfully", async () => {
+    const { stderr, exitCode } = await runCommandE2e(SKILL_ROUTES, ["skill", "init", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/bailian/i);
+  });
+});
+
+// Local-only cases: auth "none" + validation happens before any network access, no gating needed
+describe("e2e: skill (local, no credentials)", () => {
+  test("skill add without --all or --name errors as usage error (2)", async () => {
+    const { stdout, stderr, exitCode } = await runCommandE2e(SKILL_ROUTES, [
+      "skill",
+      "add",
+      "--quiet",
+    ]);
+    expect(exitCode).toBe(2);
+    expect(`${stdout}\n${stderr}`).toMatch(/--all|--name|Usage:/i);
+  });
+
+  test("skill add with both --all and --name errors as usage error (2)", async () => {
+    const { stdout, stderr, exitCode } = await runCommandE2e(SKILL_ROUTES, [
+      "skill",
+      "add",
+      "--all",
+      "--name",
+      "spark-video",
+      "--quiet",
+    ]);
+    expect(exitCode).toBe(2);
+    expect(`${stdout}\n${stderr}`).toMatch(/--all|--name|either/i);
+  });
+
+  test("skill remove without --name errors as usage error (2)", async () => {
+    const { stdout, stderr, exitCode } = await runCommandE2e(SKILL_ROUTES, [
+      "skill",
+      "remove",
+      "--quiet",
+    ]);
+    expect(exitCode).toBe(2);
+    expect(`${stdout}\n${stderr}`).toMatch(/--name|Usage:/i);
+  });
+
+  test("skill remove of a not-installed skill fails with reason (1)", async () => {
+    const configDir = makeTempConfigDir();
+    const { stdout, exitCode } = await runCommandE2e(
+      SKILL_ROUTES,
+      ["skill", "remove", "--name", "definitely-not-installed", "--output", "json"],
+      { BAILIAN_CONFIG_DIR: configDir },
+    );
+    expect(exitCode).toBe(1);
+    const data = parseStdoutJson<{
+      skills?: Array<{ name?: string; status?: string; reason?: string }>;
+    }>(stdout);
+    expect(data.skills?.[0]?.status).toBe("failed");
+    expect(data.skills?.[0]?.reason).toMatch(/not installed/i);
+  });
+});
+
+describe.skipIf(!isBailianE2EEnabled())("e2e: skill (real registry)", () => {
+  test("skill list --output json returns registry and status rows", async () => {
+    const configDir = makeTempConfigDir();
+    const { stdout, stderr, exitCode } = await runCommandE2e(
+      SKILL_ROUTES,
+      ["skill", "list", "--output", "json"],
+      { BAILIAN_CONFIG_DIR: configDir },
+    );
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      registry?: string;
+      skills?: Array<{ name?: string; status?: string }>;
+    }>(stdout);
+    expect(data.registry).toMatch(/^https?:\/\//);
+    expect(Array.isArray(data.skills)).toBe(true);
+  }, 60_000);
+
+  test("skill add + remove full lifecycle in isolated dirs", async () => {
+    const configDir = makeTempConfigDir();
+    // Empty fake home → no agents detected → fan-out never leaves the sandbox
+    const fakeHome = makeTempConfigDir();
+    const env = { BAILIAN_CONFIG_DIR: configDir, HOME: fakeHome, USERPROFILE: fakeHome };
+
+    const added = await runCommandE2e(
+      SKILL_ROUTES,
+      ["skill", "add", "--name", WIKI_SKILL, "--output", "json"],
+      env,
+    );
+    expect(added.exitCode, added.stderr).toBe(0);
+    const addData = parseStdoutJson<{ skills?: Array<{ name?: string; status?: string }> }>(
+      added.stdout,
+    );
+    expect(addData.skills?.[0]?.status).toBe("installed");
+    expect(existsSync(join(configDir, "skills", WIKI_SKILL, "SKILL.md"))).toBe(true);
+
+    const removed = await runCommandE2e(
+      SKILL_ROUTES,
+      ["skill", "remove", "--name", WIKI_SKILL, "--output", "json"],
+      env,
+    );
+    expect(removed.exitCode, removed.stderr).toBe(0);
+    const removeData = parseStdoutJson<{ skills?: Array<{ name?: string; status?: string }> }>(
+      removed.stdout,
+    );
+    expect(removeData.skills?.[0]?.status).toBe("removed");
+    expect(existsSync(join(configDir, "skills", WIKI_SKILL))).toBe(false);
+  }, 300_000);
+});
@@ -20,7 +20,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: text chat（DashScope）", () => {
      "text",
      "chat",
      "--model",
-      "qwen3.7-max",
+      "qwen3.8-max",
    ]);
    expect(exitCode).toBe(2);
    expect(stderr).toMatch(/--message|Usage:/i);
@@ -32,7 +32,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: text chat（DashScope）", () => {
      "chat",
      "--dry-run",
      "--model",
-      "qwen3.7-max",
+      "qwen3.8-max",
      "--message",
      "干跑",
      "--max-tokens",
@@ -44,16 +44,16 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: text chat（DashScope）", () => {
    const data = parseStdoutJson<{
      request?: { model?: string; messages?: Array<{ content?: string }> };
    }>(stdout);
-    expect(data.request?.model).toBe("qwen3.7-max");
+    expect(data.request?.model).toBe("qwen3.8-max");
    expect(data.request?.messages?.some((m) => m.content === "干跑")).toBe(true);
  });

-  test("【qwen3.7-max】文本对话", async () => {
+  test("【qwen3.8-max】文本对话", async () => {
    const { stdout, stderr, exitCode } = await runCommandE2e(TEXT_CHAT_ROUTES, [
      "text",
      "chat",
      "--model",
-      "qwen3.7-max",
+      "qwen3.8-max",
      "--message",
      "只回复一个字：好",
      "--max-tokens",
@@ -10,6 +10,10 @@ export const AUTH_ROUTES: E2eRouteExports = {
  "auth logout": "authLogout",
 };

+export const UPDATE_ROUTES: E2eRouteExports = {
+  update: "update",
+};
+
 export const TEXT_CHAT_ROUTES: E2eRouteExports = { "text chat": "textChat" };

 export const CONFIG_ROUTES: E2eRouteExports = {
@@ -29,6 +33,10 @@ export const MEMORY_ROUTES: E2eRouteExports = {
  "memory delete": "memoryDelete",
  "memory profile create": "memoryProfileCreate",
  "memory profile get": "memoryProfileGet",
+  "memory profile list": "memoryProfileList",
+  "memory profile detail": "memoryProfileDetail",
+  "memory profile update": "memoryProfileUpdate",
+  "memory profile delete": "memoryProfileDelete",
 };

 export const KNOWLEDGE_ROUTES: E2eRouteExports = {
@@ -156,11 +164,20 @@ export const TOKEN_PLAN_ROUTES: E2eRouteExports = {
  "token-plan add-member": "tokenPlanAddMember",
 };

+export const SKILL_ROUTES: E2eRouteExports = {
+  "skill add": "skillAdd",
+  "skill update": "skillUpdate",
+  "skill remove": "skillRemove",
+  "skill list": "skillList",
+  "skill init": "skillInit",
+};
+
 export const MANAGED_AGENT_ROUTES: E2eRouteExports = {
  "managed-agent init": "managedAgentInit",
  "managed-agent validate": "managedAgentValidate",
  "managed-agent plan": "managedAgentPlan",
  "managed-agent apply": "managedAgentApply",
+  "managed-agent run": "managedAgentRun",
  "managed-agent destroy": "managedAgentDestroy",
  "managed-agent state list": "managedAgentStateList",
  "managed-agent state rm": "managedAgentStateRm",
@@ -0,0 +1,33 @@
+import { describe, expect, test } from "vite-plus/test";
+import { runCommandE2e } from "./helpers.ts";
+import { UPDATE_ROUTES } from "./topic-routes.ts";
+
+describe("e2e: update", () => {
+  test("update --help 正常退出并展示 --to", async () => {
+    const { stderr, exitCode } = await runCommandE2e(UPDATE_ROUTES, ["update", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--to/);
+    expect(stderr).toMatch(/<version>/);
+  });
+
+  test("update --help 包含 --to 示例", async () => {
+    const { stderr, exitCode } = await runCommandE2e(UPDATE_ROUTES, ["update", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("--to 0.1.14");
+  });
+
+  test("update --to 缺值时退出为用法错误 (2)", async () => {
+    const { stderr, exitCode } = await runCommandE2e(UPDATE_ROUTES, ["update", "--to"]);
+    expect(exitCode, stderr).toBe(2);
+  });
+
+  test("update --to 非法版本时退出为用法错误 (2)", async () => {
+    const { stderr, exitCode } = await runCommandE2e(UPDATE_ROUTES, [
+      "update",
+      "--to",
+      "not-a-version",
+    ]);
+    expect(exitCode, stderr).toBe(2);
+    expect(stderr).toMatch(/semver|--to/i);
+  });
+});
@@ -159,7 +159,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
        "image",
        "generate",
        "--model",
-        "qwen-image-2.0",
+        "qwen-image-3.0",
        "--prompt",
        "一只简笔画小猫，白底",
        "--out-dir",
@@ -169,7 +169,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
        "image",
        "generate",
        "--model",
-        "qwen-image-2.0",
+        "qwen-image-3.0",
        "--prompt",
        "一片绿色的树叶，白底",
        "--out-dir",
@@ -13,7 +13,7 @@ describe("e2e: vision describe", () => {
        "token-plan": {
          api_key: "sk-sp-e2e-placeholder",
          base_url: "https://token-plan.cn-beijing.maas.aliyuncs.com",
-          default_text_model: "qwen3.8-max-preview",
+          default_text_model: "qwen3.8-max",
        },
      }),
    );
@@ -40,6 +40,6 @@ describe("e2e: vision describe", () => {

    expect(exitCode, stderr).toBe(0);
    const data = parseStdoutJson<{ request?: { model?: string } }>(stdout);
-    expect(data.request?.model).toBe("qwen3.8-max-preview");
+    expect(data.request?.model).toBe("qwen3.8-max");
  });
 });
@@ -1,6 +1,6 @@
 {
  "name": "bailian-cli-core",
-  "version": "1.13.0",
+  "version": "1.14.2",
  "description": "Core SDK for bailian-cli. See https://www.npmjs.com/package/bailian-cli for usage.",
  "homepage": "https://bailian.console.aliyun.com/cli",
  "bugs": {
@@ -40,11 +40,13 @@
    "check": "vp check"
  },
  "dependencies": {
+    "tar-stream": "catalog:",
    "yaml": "^2.8.3",
    "yauzl": "catalog:"
  },
  "devDependencies": {
    "@types/node": "catalog:",
+    "@types/tar-stream": "catalog:",
    "@types/yauzl": "catalog:",
    "@typescript/native-preview": "7.0.0-dev.20260328.1",
    "typescript": "^6.0.2",
@@ -7,6 +7,7 @@ export { recallCandidates } from "./recall.ts";
 export { recallSemantic, isSemanticAvailable } from "./recall-semantic.ts";
 export type { RecommendOptions } from "./recommend.ts";
 export { buildDocLink, rankModels } from "./recommend.ts";
+export { maybeSyncWikiData } from "./sync.ts";
 export type { ModelSource } from "./sources/types.ts";
 export type {
  Budget,
@@ -1,6 +1,5 @@
-import { cpSync, existsSync, mkdirSync, readFileSync } from "node:fs";
-import { dirname, join } from "node:path";
-import { fileURLToPath } from "node:url";
+import { existsSync, readFileSync } from "node:fs";
+import { join } from "node:path";
 import { getConfigDir } from "../../config/paths.ts";
 import type { ModelPrice, ModelProfile, QpmLimit } from "../types.ts";
 import type { ModelSource } from "./types.ts";
@@ -13,12 +12,9 @@ function getCatalogDir(): string {
 }

 function getCatalogPath(): string {
-  return join(getCatalogDir(), MODELS_FILE);
-}
-
-function getMonorepoModelsDir(): string {
-  const coreDir = dirname(fileURLToPath(import.meta.url));
-  return join(coreDir, "../../../../../skills/bailian-docs-llm-wiki/models");
+  // Full-package layout keeps the `models/` subdir (raw/, wiki/, models/, …),
+  // so models.jsonl lives at <skill>/models/models.jsonl — not at the skill root.
+  return join(getCatalogDir(), "models", MODELS_FILE);
 }

 function fromJsonlRecord(raw: Record<string, unknown>): ModelProfile | null {
@@ -62,41 +58,24 @@ function readJsonlModels(filePath: string): ModelProfile[] {
  return models;
 }

-function installFromMonorepo(): boolean {
-  const src = getMonorepoModelsDir();
-  if (!existsSync(join(src, MODELS_FILE))) return false;
-  const dest = getCatalogDir();
-  try {
-    mkdirSync(dest, { recursive: true });
-    cpSync(src, dest, { recursive: true });
-    return true;
-  } catch {
-    return false;
-  }
-}
-
 export interface CatalogSourceOptions {
  onPrepareStart?: () => void;
 }

 export class CatalogSource implements ModelSource {
  readonly name = "catalog";
-  private options: CatalogSourceOptions;

-  constructor(options?: CatalogSourceOptions) {
-    this.options = options ?? {};
-  }
+  // Options retained for API compatibility. Data is now always provisioned by
+  // the CLI postinstall hook and refreshed by advisor sync, so the previous
+  // `onPrepareStart` install callback is obsolete.
+  constructor(_options?: CatalogSourceOptions) {}

  available(): boolean {
    return existsSync(getCatalogPath());
  }

  async load(): Promise<ModelProfile[]> {
-    if (!this.available()) {
-      this.options.onPrepareStart?.();
-      const installed = installFromMonorepo();
-      if (!installed) return [];
-    }
+    if (!this.available()) return [];
    return readJsonlModels(getCatalogPath());
  }
 }
@@ -0,0 +1,173 @@
+/**
+ * sync.ts — Wiki data sync (layer 2: triggered by recommend)
+ *
+ * Called via `maybeSyncWikiData()` during `bl advisor recommend`:
+ *   1. 12h throttle: skip if last check was less than 12h ago
+ *   2. Download skills/index.json from public-read OSS, compare bailian-docs-llm-wiki entry version
+ *   3. Same version → only refresh lastChecked
+ *   4. Different version → delegate to the shared skill install pipeline
+ *      (installSkill: download + extract + SKILL.md validate + atomic swap;
+ *       linkSkillToAgents: fan-out symlinks to detected agents;
+ *       upsertSkillLockEntry: write lock WITH links so bl skill remove can reclaim correctly)
+ *
+ * Protocol: unified skill publishing protocol (FC publish-skills, all skills are isomorphic), entry point is
+ * skills/index.json, one content-addressed object per skill (sha256-<hex>.tar.br, brotli q6;
+ * legacy fallback skill.tar.br).
+ *
+ * Complements postinstall.js (layer 1, unconditional overwrite on npm install). Install, extraction,
+ * validation, fan-out and lock writing all reuse the skills/ module (same as bl skill add), symmetric
+ * with publisher tar.pack().
+ *
+ * Failure strategy: any step failure silently returns without updating lastChecked; next recommend retries immediately.
+ */
+import { existsSync, readFileSync, writeFileSync } from "node:fs";
+import { join } from "node:path";
+import { getConfigDir } from "../config/paths.ts";
+import { detectInstalledAgents, fanOutSkillToAgents } from "../skills/agents.ts";
+import { buildSkillLockEntry, installSkillWithFanout } from "../skills/installer.ts";
+import { readSkillLock, upsertSkillLockEntry } from "../skills/lock.ts";
+import { fetchSkillsIndex } from "../skills/registry.ts";
+import type { SkillIndexEntry, SkillLockEntry } from "../skills/types.ts";
+
+const WIKI_SKILL_NAME = "bailian-docs-llm-wiki";
+const SKILL_DIR_NAME = "skills/bailian-docs-llm-wiki";
+const STATE_FILE_NAME = "wiki-sync-state.json";
+const MODELS_FILE = "models.jsonl";
+
+const THROTTLE_MS = 12 * 60 * 60 * 1000; // 12h
+/** Tighter than the interactive default: the silent channel must not stall `bl advisor recommend` */
+const INDEX_TIMEOUT_MS = 3000;
+
+interface SyncState {
+  lastChecked: number;
+  /** Content fingerprint of the last synced revision; the change-detection token */
+  contentHash: string;
+}
+
+function getCatalogDir(): string {
+  return join(getConfigDir(), SKILL_DIR_NAME);
+}
+
+/**
+ * Whether local Wiki data is ready. Uses `models.jsonl` as the existence signal, consistent with
+ * `CatalogSource.available()`: as long as the file advisor actually consumes exists,
+ * the data is considered available.
+ */
+function catalogDataExists(): boolean {
+  return existsSync(join(getCatalogDir(), "models", MODELS_FILE));
+}
+
+function getStatePath(): string {
+  return join(getConfigDir(), STATE_FILE_NAME);
+}
+
+function readState(): SyncState | null {
+  try {
+    return JSON.parse(readFileSync(getStatePath(), "utf-8")) as SyncState;
+  } catch {
+    return null;
+  }
+}
+
+function writeState(state: SyncState): void {
+  try {
+    writeFileSync(getStatePath(), JSON.stringify(state));
+  } catch {
+    /* Non-critical: if state write fails, next run will re-check */
+  }
+}
+
+/**
+ * Record this sync in skill-lock.json so the wiki skill shares the same ledger as bl skill
+ * (list shows installed instead of untracked; update/remove can manage it correctly).
+ * Includes fan-out links so bl skill remove can reclaim agent symlinks.
+ * Bookkeeping in the silent channel must be best-effort: failure does not affect sync results.
+ */
+function recordWikiInLock(lockEntry: SkillLockEntry): void {
+  try {
+    upsertSkillLockEntry(WIKI_SKILL_NAME, lockEntry);
+  } catch {
+    /* Bookkeeping failure does not block sync; next sync or bl skill add will fill it in */
+  }
+}
+
+/**
+ * Whether the lock still needs a wiki backfill: content fingerprint mismatch, or the
+ * record carries no fan-out links (postinstall writes contentHash only and never fans
+ * out, so agents would otherwise never see the wiki skill until content changes).
+ */
+function wikiLockNeedsBackfill(contentHash: string): boolean {
+  try {
+    const locked = readSkillLock().skills[WIKI_SKILL_NAME];
+    return locked?.contentHash !== contentHash || !Array.isArray(locked.links);
+  } catch {
+    return true;
+  }
+}
+
+/** Fetch skills/index.json via the shared registry client and extract the wiki skill entry; returns null on any failure */
+async function fetchIndexEntry(): Promise<SkillIndexEntry | null> {
+  try {
+    const index = await fetchSkillsIndex(INDEX_TIMEOUT_MS);
+    return index.skills[WIKI_SKILL_NAME] ?? null;
+  } catch {
+    return null;
+  }
+}
+
+/**
+ * Check and sync Wiki data. Runs silently; never throws.
+ * @returns Whether data was actually updated (for testing/debugging)
+ */
+export async function maybeSyncWikiData(): Promise<boolean> {
+  const state = readState();
+  const now = Date.now();
+
+  // 1. throttle gate: only skip when "within the 12h window" AND "local data actually exists".
+  //    If data is missing (user deleted manually, postinstall failed but state remains, etc.),
+  //    ignore throttle and sync immediately to ensure advisor has data.
+  if (state && now - state.lastChecked < THROTTLE_MS && catalogDataExists()) {
+    return false;
+  }
+
+  // 2. Fetch skills/index.json and get the wiki entry
+  const entry = await fetchIndexEntry();
+  if (!entry?.contentHash) return false; // On failure, do not write lastChecked; retry next time
+
+  // 3. Same content and local data exists: only refresh lastChecked, no re-download needed.
+  //    Covers two cases: (a) state.contentHash === entry.contentHash → direct hit;
+  //    (b) state missing but data intact (user or accident only deleted state) → write the fingerprint
+  //    back to state, avoiding unnecessary download+extract.
+  //    If data is missing or the fingerprint differs, falls through to step 4 for full download.
+  const dataOk = catalogDataExists();
+  if (dataOk && (!state || state.contentHash === entry.contentHash)) {
+    writeState({ lastChecked: now, contentHash: entry.contentHash });
+    // Lock record missing/stale (e.g. postinstall wrote canonical only, without fan-out) → backfill
+    if (wikiLockNeedsBackfill(entry.contentHash)) {
+      const previousLinks = readSkillLock().skills[WIKI_SKILL_NAME]?.links ?? [];
+      const fanout = fanOutSkillToAgents(WIKI_SKILL_NAME, detectInstalledAgents(), previousLinks);
+      recordWikiInLock(buildSkillLockEntry(entry, fanout.links));
+    }
+    return false;
+  }
+
+  // 4. Different content or missing data: delegate to the shared skill install pipeline
+  //    (download → extract → SKILL.md validate → atomic swap → fan-out → lock with links)
+  try {
+    const previousLinks = readSkillLock().skills[WIKI_SKILL_NAME]?.links ?? [];
+    const record = await installSkillWithFanout(
+      WIKI_SKILL_NAME,
+      entry,
+      detectInstalledAgents(),
+      previousLinks,
+    );
+    recordWikiInLock(record.lockEntry);
+  } catch {
+    // Install failed → clean exit, leave existing data untouched, do not write state; next recommend retries
+    return false;
+  }
+
+  // 5. Success: write state
+  writeState({ lastChecked: now, contentHash: entry.contentHash });
+  return true;
+}
@@ -126,7 +126,10 @@ export class Client {
  /** Resolve a file arg: upload a local path to OSS (returns oss:// URL), or pass a URL through. */
  uploadFile(source: string, model: string, opts: { signal?: AbortSignal } = {}): Promise<string> {
    if (!isLocalFile(source)) return Promise.resolve(source);
-    return resolveFileUrl(source, this.requireApi().token, model, opts);
+    return resolveFileUrl(source, this.requireApi().token, model, {
+      ...opts,
+      identity: this.deps.identity,
+    });
  }

  /**
@@ -233,7 +236,7 @@ export class Client {
    const timeoutMs = this.deps.settings.timeout * 1000;
    const res = await fetch(endpoint, {
      method: opts.method,
-      headers: { ...headers, ...trackingHeaders() },
+      headers: { ...headers, ...trackingHeaders(this.deps.identity) },
      body: bodyStr || undefined,
      signal: AbortSignal.timeout(timeoutMs),
    });
@@ -75,7 +75,11 @@ export function profileSchemaPath(): string {
 }

 export function userProfilePath(schemaId: string): string {
-  return `/api/v2/apps/memory/profile_schemas/${encodeURIComponent(schemaId)}/profiles`;
+  return `/api/v2/apps/memory/profile_schemas/${encodeURIComponent(schemaId)}/user_profile`;
+}
+
+export function profileSchemaItemPath(schemaId: string): string {
+  return `/api/v2/apps/memory/profile_schemas/${encodeURIComponent(schemaId)}`;
 }

 // ---- Knowledge Base Retrieve (DashScope) ----
@@ -1,23 +1,31 @@
 /**
 * Shared HTTP request headers for all outgoing requests.
 *
- * Centralises the `x-dashscope-source-config` header so every fetch call
- * (both via the central http client and the bypass paths) uses the
- * same values from a single source of truth.
+ * Centralises the `x-dashscope-source-config` header so Bailian/DashScope API
+ * transports use the same product identity. Generic npm, OSS, and result-file
+ * transfers deliberately do not send this gateway-consumed metadata.
 */

+import type { Identity } from "../config/schema.ts";
+
 export const CHANNEL = "bailian-cli";

-export const TAGS = { t1: "public", t2: "" };
+export type TrackingIdentity = Pick<Identity, "binName" | "version">;

-export const SOURCE_CONFIG = JSON.stringify({
-  channel: CHANNEL,
-  tags: TAGS,
-});
+export function sourceConfig(identity: TrackingIdentity): string {
+  return JSON.stringify({
+    channel: CHANNEL,
+    tags: {
+      t1: "public",
+      t2: identity.binName,
+      t3: identity.version,
+    },
+  });
+}

-/** Standard tracking headers required on every outbound request. */
-export function trackingHeaders(): Record<string, string> {
+/** Tracking headers for Bailian/DashScope API requests. */
+export function trackingHeaders(identity: TrackingIdentity): Record<string, string> {
  return {
-    "x-dashscope-source-config": SOURCE_CONFIG,
+    "x-dashscope-source-config": sourceConfig(identity),
  };
 }
@@ -4,7 +4,7 @@ import { BailianError } from "../errors/base.ts";
 import { ExitCode } from "../errors/codes.ts";
 import { mapApiError } from "../errors/api.ts";
 import { maskToken } from "../utils/token.ts";
-import { SOURCE_CONFIG, trackingHeaders } from "./headers.ts";
+import { sourceConfig, trackingHeaders } from "./headers.ts";

 /** 传输层依赖:UA 用 identity,timeout/verbose 用 settings。凭证由调用方(Client)注头。 */
 export interface HttpDeps {
@@ -39,7 +39,7 @@ export async function request(deps: HttpDeps, opts: RequestOpts): Promise<Respon

  const headers: Record<string, string> = {
    "User-Agent": `${deps.identity.clientName}/${deps.identity.version}`,
-    ...trackingHeaders(),
+    ...trackingHeaders(deps.identity),
    ...opts.headers,
  };

@@ -59,7 +59,7 @@ export async function request(deps: HttpDeps, opts: RequestOpts): Promise<Respon
    console.error(`> ${opts.method ?? "GET"} ${opts.url}`);
    const auth = headers["Authorization"];
    if (auth) console.error(`> Auth: ${maskToken(auth.replace(/^Bearer /, ""))}`);
-    console.error(`> x-dashscope-source-config: ${SOURCE_CONFIG}`);
+    console.error(`> x-dashscope-source-config: ${sourceConfig(deps.identity)}`);
  }

  const timeoutMs = (opts.timeout ?? deps.settings.timeout) * 1000;
@@ -10,7 +10,7 @@ import { image2ImagePath, imagePath, imageSyncPath, imageText2ImagePath } from "
 * - async text2image + prompt: wan2.5/2.2/2.1-t2i*, wanx*-t2i*
 *
 * Edit (I2I):
- * - sync multimodal + messages(+images): qwen-image-2.0*, qwen-image-edit*, wan2.6-image*, wan2.7-image*
+ * - sync multimodal + messages(+images): qwen-image-3.0*, qwen-image-2.0*, qwen-image-edit*, wan2.6-image*, wan2.7-image*
 *   (pure T2I models such as z-image / qwen-image-plus / qwen-image-max are NOT edit models)
 * - async image2image + prompt/images: wan2.5-i2i*
 * - async image2image + function/base_image_url: *imageedit* (e.g. wanx2.1-imageedit)
@@ -60,6 +60,7 @@ const SYNC_GENERATE_PREFIXES = ["qwen-image", "wan2.7-image", "z-image"] as cons
 * Pure T2I models (z-image / qwen-image-plus / qwen-image-max) are excluded.
 */
 const SYNC_EDIT_PREFIXES = [
+  "qwen-image-3.0",
  "qwen-image-2.0",
  "qwen-image-edit",
  "wan2.7-image",
@@ -109,7 +110,12 @@ export function isWanxFunctionImageEditModel(model: string): boolean {
 }

 export function resolveImageSizeProfile(model: string): ImageSizeProfile {
-  if (model.startsWith("qwen-image-2.0") || model.startsWith("qwen-image-edit")) {
+  // 3.0 暂复用 2.0 高分比例表（CLI ratio→像素便捷映射）；不做独立 3.0 profile。
+  if (
+    model.startsWith("qwen-image-3.0") ||
+    model.startsWith("qwen-image-2.0") ||
+    model.startsWith("qwen-image-edit")
+  ) {
    return "qwen-image-2.0";
  }
  // Remaining qwen-image* (plus / max / bare qwen-image) share the fixed table.
@@ -133,7 +139,13 @@ export function resolveImageSizeProfile(model: string): ImageSizeProfile {

 /** Official / CLI defaults for prompt_extend when the flag is omitted. */
 export function resolvePromptExtendDefault(model: string): boolean | undefined {
-  if (model.startsWith("qwen-image-2.0") || model.startsWith("qwen-image-max")) return true;
+  if (
+    model.startsWith("qwen-image-3.0") ||
+    model.startsWith("qwen-image-2.0") ||
+    model.startsWith("qwen-image-max")
+  ) {
+    return true;
+  }
  // Z-Image docs default prompt_extend to false.
  if (model.startsWith("z-image")) return false;
  return undefined;
@@ -13,6 +13,7 @@ export {
  memoryNodePath,
  memorySearchPath,
  mcpWebSearchPath,
+  profileSchemaItemPath,
  profileSchemaPath,
  speechRecognizePath,
  speechSynthesizePath,
@@ -34,7 +35,7 @@ export {
  type ImageInputStyle,
  type ImageSizeProfile,
 } from "./image-routes.ts";
-export { CHANNEL, SOURCE_CONFIG, TAGS, trackingHeaders } from "./headers.ts";
+export { CHANNEL, sourceConfig, trackingHeaders, type TrackingIdentity } from "./headers.ts";
 export type { HttpDeps, RequestOpts } from "./http.ts";
 export { request, requestJson } from "./http.ts";
 export { createInstrumentedFetch, type FetchImplementation } from "./instrumented-fetch.ts";
@@ -50,7 +50,7 @@ export function createInstrumentedFetch(deps: HttpDeps): FetchImplementation {
      headers.set("User-Agent", `${deps.identity.clientName}/${deps.identity.version}`);
    }
    if (isAlibabaCloudHost(url)) {
-      for (const [name, value] of Object.entries(trackingHeaders())) {
+      for (const [name, value] of Object.entries(trackingHeaders(deps.identity))) {
        headers.set(name, value);
      }
    }
@@ -148,7 +148,7 @@ export class McpClient {
      "Content-Type": "application/json",
      Accept: "application/json, text/event-stream",
      "User-Agent": `${this.deps.identity.clientName}/${this.deps.identity.version}`,
-      ...trackingHeaders(),
+      ...trackingHeaders(this.deps.identity),
    };

    if (this.authToken) {
@@ -10,7 +10,7 @@ interface ModelProfilePreset {
 const MODEL_PROFILE_PRESETS: Readonly<Record<string, ModelProfilePreset>> = {
  "token-plan": {
    baseUrl: "https://token-plan.cn-beijing.maas.aliyuncs.com",
-    defaultTextModel: "qwen3.8-max-preview",
+    defaultTextModel: "qwen3.8-max",
    defaultVideoModel: "happyhorse-1.1-t2v",
    defaultImageToVideoModel: "happyhorse-1.1-i2v",
    defaultReferenceToVideoModel: "happyhorse-1.1-r2v",
@@ -47,7 +47,7 @@ export interface DatasetUploadParams {
 export async function uploadDataset(
  client: Client,
  params: DatasetUploadParams,
-): Promise<DatasetFile> {
+): Promise<DatasetFile & { request_id?: string }> {
  const { filePath, purpose = "fine-tune", signal } = params;
  const stat = statSync(filePath);
  const fileName = basename(filePath);
@@ -75,6 +75,7 @@ export async function uploadDataset(
      size: body.bytes ?? stat.size,
      purpose: body.purpose ?? purpose,
      gmt_create: body.created_at ? new Date(body.created_at * 1000).toISOString() : undefined,
+      request_id: body.request_id,
    };
  }
  // No id in response → upload reported HTTP 200 but produced no usable record
@@ -9,7 +9,7 @@ import { existsSync, readFileSync, statSync } from "fs";
 import { basename, extname } from "path";
 import { BailianError } from "../errors/base.ts";
 import { ExitCode } from "../errors/codes.ts";
-import { trackingHeaders } from "../client/headers.ts";
+import { trackingHeaders, type TrackingIdentity } from "../client/headers.ts";
 import { REGIONS } from "../config/schema.ts";

 // Pinned to cn region; thread baseUrl through if overseas upload becomes a requirement.
@@ -36,6 +36,7 @@ interface UploadPolicyResponse {
 async function getUploadPolicy(
  apiKey: string,
  model: string,
+  identity: TrackingIdentity,
  signal?: AbortSignal,
 ): Promise<UploadPolicy> {
  const url = `${UPLOAD_API}?action=getPolicy&model=${encodeURIComponent(model)}`;
@@ -44,7 +45,7 @@ async function getUploadPolicy(
    headers: {
      Authorization: `Bearer ${apiKey}`,
      "Content-Type": "application/json",
-      ...trackingHeaders(),
+      ...trackingHeaders(identity),
    },
    signal: policySignal.signal,
  }).finally(policySignal.cleanup);
@@ -87,9 +88,6 @@ async function uploadToOSS(
  const uploadSignal = combineWithTimeout(120_000, signal);
  const res = await fetch(policy.upload_host, {
    method: "POST",
-    headers: {
-      ...trackingHeaders(),
-    },
    body: form,
    signal: uploadSignal.signal,
  }).finally(uploadSignal.cleanup);
@@ -109,6 +107,7 @@ export interface UploadOptions {
  apiKey: string;
  model: string;
  filePath: string;
+  identity: TrackingIdentity;
  signal?: AbortSignal;
 }

@@ -160,7 +159,7 @@ export function redactDataUri(input: string): string {
 * The URL is valid for 48 hours.
 */
 export async function uploadFile(opts: UploadOptions): Promise<string> {
-  const { apiKey, model, filePath, signal } = opts;
+  const { apiKey, model, filePath, identity, signal } = opts;

  if (!existsSync(filePath)) {
    throw new BailianError(`File not found: ${filePath}`, ExitCode.USAGE);
@@ -171,7 +170,7 @@ export async function uploadFile(opts: UploadOptions): Promise<string> {
    throw new BailianError(`Not a file: ${filePath}`, ExitCode.USAGE);
  }

-  const policy = await getUploadPolicy(apiKey, model, signal);
+  const policy = await getUploadPolicy(apiKey, model, identity, signal);
  return uploadToOSS(policy, filePath, signal);
 }

@@ -193,10 +192,16 @@ export async function resolveFileUrl(
  input: string,
  apiKey: string,
  model: string,
-  opts: { signal?: AbortSignal } = {},
+  opts: { identity: TrackingIdentity; signal?: AbortSignal },
 ): Promise<string> {
  if (!isLocalFile(input)) return input;
-  return uploadFile({ apiKey, model, filePath: input, signal: opts.signal });
+  return uploadFile({
+    apiKey,
+    model,
+    filePath: input,
+    identity: opts.identity,
+    signal: opts.signal,
+  });
 }

 function combineWithTimeout(
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
lisheng.lisheng	f919ebae3c	feat(dsh): remote managed-agent as on-demand tool + bl managed-agent run Rework the managed-agent integration so a dsh user can, in plain language, have a Bailian cloud agent created and run a task — no hand-written agents.yaml, no prior apply. New `bl managed-agent run --prompt <task> [--instructions] [--model] [--agent]`: one step that idempotently materializes a cloud agent + its environment, then opens a session and streams the result. It mirrors the OpenAgentPack webui backend's ensure+run recipe (resolveProjectConfigFrom Object → syncAgentResourcesWithStateBackend → readProjectRuntime + startSessionRun) from an in-memory config, reusing the existing credential spine in _engine/credentials.ts. State persists under the bl config dir (~/.bailian/managed-agent/<agent>/), never the user's cwd, so repeat runs with the same --agent reuse the materialized agent. Unlike apply it provisions without --yes, since running is the intent. dsh side: replace the SubagentProvider with a plain tool `bailian_run_remote_task` (packages/dsh/src/tool-managed-agent). The subagent seam did not fit: in the web profile every tool-subagent row is disabled in the host plane (delegation lives in agent presets), a provider fixes one agent identity in config, and the default numeric maxDepth would fail-mount a no-depthLimit provider. As a tool the model calls it directly and fills `instructions` from the user's intent, so the remote agent's role is defined per task. Enabled by default — it creates nothing at load, only on invocation. LLM row: configure the base bundle's existing llm-pi-ai row instead of mounting a second pi-ai instance (a second instance re-declares pi-ai's global configurable-provider catalog and fails boot on a duplicate amazon-bedrock). TokenPlan reads a dedicated BAILIAN_TOKENPLAN_API_KEY, not DASHSCOPE_API_KEY: TokenPlan (sk-sp-) and pay-as-you-go (sk-ws-) keys 401 each other's endpoints, so sharing one var would silently break whichever plugin lost. Note: the ensure+run happy path could not be verified end-to-end on the available account — agentstudio returns 404 there, and the existing `managed-agent apply` 404s identically against the same endpoint/key, so the failure is account/service provisioning, not this change. Command wiring, dry-run, config assembly, credential injection and URL construction were all verified. Co-Authored-By: Claude <noreply@anthropic.com>	2026-08-15 10:29:24 +08:00
lisheng.lisheng	9e9911aa86	feat(dsh): add bailian-cli-dsh plugin bundle for DeepSeek Harness Expose Bailian capabilities to dsh through its service seams as one package with six subpath plugin entries and a `dsh.bundle` patch: - TokenPlan as an LLM provider - bailian_vision_describe / bailian_image_generate tools over `bl` - knowledge-base retrieval as a WebSearchProvider (`bailian-kb`) - cross-session memory: tools, pre-step recall, turn-close persist - managed-agent as a SubagentProvider TokenPlan configures the base bundle's existing pi-ai row rather than mounting a second `dsh-llm-pi-ai` instance. A second instance cannot work: pi-ai re-declares its entire built-in provider catalog to `registerConfigurableProviders`, and that directory is global, so boot fails with a duplicate on `amazon-bedrock`. Routes and vision support were probed against the live gateway. qwen3.8-max, qwen3.7-plus, qwen3.6-flash and glm-5.2 read images; qwen3.7-max rejects them with HTTP 400; the DeepSeek routes accept image content without erroring yet stay blind. The DeepSeek entries therefore do not declare image input — claiming it would turn a clean refusal into a silently wrong answer — and `bailian_vision_describe` serves them by returning text instead. Also fix `bl memory` against the v2 API, each verified live: - `profile get` used /profiles, which returns HTTP 500. The documented and working endpoint is /user_profile. - `add` read `response.memory_ids`, which the service never returns. It returns `memory_nodes`, so text output always printed "IDs: none". - `MemoryNode.created_at`/`updated_at` are unix seconds, not strings, and `UserProfileResponse.profile` did not match the wire shape. - Add the missing request parameters: --meta-data, --project-id, --project-ids, --min-score, --enable-rerank, --plan-version, --enable-judge, --enable-rewrite, --timestamp. - Add `memory profile list\|detail\|update\|delete`, covering the four v2 profile-schema operations the CLI was missing. `plan_version: lite` is ignored by the service and still bills pro; `enable_rerank: false` is what actually selects lite, which is ~50x cheaper per search. The CLI flag and the memory plugin both send the parameter that works. Disable pnpm's autoInstallPeers: the @deepseek-ai/dsh-* rc line peers on three packages that were never published to npm, which 404s the whole workspace install. Verified the existing packages still build. Co-Authored-By: Claude <noreply@anthropic.com>	2026-08-15 01:47:18 +08:00
gujieye	2389681ad6	Merge pull request #144 from modelstudioai/feat/add-version-tag feat: add version 1.14.2	2026-08-07 18:02:08 +08:00
故璃	946b7029c6	feat: add version 1.14.2	2026-08-07 17:42:31 +08:00
gujieye	b9ecd5c43b	Merge pull request #143 from modelstudioai/feat/skill-init-commend feat: add skill init & opt commend flags	2026-08-07 17:26:21 +08:00
Gong Shiqi	978f332fea	Merge pull request #142 from modelstudioai/docs/update-readme-and-agent-guides docs: refresh READMEs and auth maintenance guidance	2026-08-07 17:24:52 +08:00
故璃	4502424200	feat: add skill init & opt commend flags	2026-08-07 17:17:27 +08:00
若麒	03839766bc	docs: update READMEs	2026-08-07 17:15:26 +08:00
若麒	1f8b9ace7e	docs: refine auth maintenance guidance	2026-08-07 15:27:46 +08:00
Gong Shiqi	6338df36be	Merge pull request #138 from modelstudioai/feat/update-defmodel Update default image model to qwen-image-3.0	2026-08-05 19:43:48 +08:00
若麒	cb6740965f	chore(release): prepare 1.14.1	2026-08-05 19:35:05 +08:00
Gong Shiqi	2dffee5b7a	Merge pull request #139 from modelstudioai/feat/source-config-tags feat: add CLI source config tags	2026-08-05 17:44:40 +08:00
若麒	01ec13aad8	feat: add CLI source config tags	2026-08-05 17:37:14 +08:00
clh02467605	b68ff45fb9	Merge remote-tracking branch 'refs/remotes/origin/main' into feat/update-defmodel	2026-08-05 17:08:44 +08:00
clh02467605	4990b27436	feat: update image default model	2026-08-05 16:58:29 +08:00
gujieye	262681484b	Merge pull request #137 from modelstudioai/feat/deploy-update feat: align agent registry with upstream and harden cross-platform install	2026-08-05 16:21:10 +08:00
故璃	8488b251f7	Merge branch 'main' into feat/deploy-update	2026-08-05 16:11:38 +08:00
Gong Shiqi	b1908fa879	Merge pull request #134 from modelstudioai/chore/opti-skill refactor(skills): split domain skills and introduce bailian-protocol companion	2026-08-05 11:16:08 +08:00
clh02467605	d64ba09bef	merge: merged main to current branch	2026-08-05 10:59:26 +08:00
clh02467605	8cdd54cf7a	docs(skills): remove companions claim; make --all -g the supported install path	2026-08-05 10:27:39 +08:00
故璃	121fa1317f	feat(skills): align agent registry with upstream and harden cross-platform install	2026-08-05 10:17:06 +08:00
Gong Shiqi	564e21d9f1	Merge pull request #130 from modelstudioai/feat/multi-channel-install Feat/multi channel install	2026-08-04 20:30:54 +08:00
若麒	081d09863b	Merge branch 'main' into feat/multi-channel-install	2026-08-04 20:22:26 +08:00
clh02467605	17b13de162	merge: merged main to current branch	2026-08-04 18:43:50 +08:00
clh02467605	ca98d8a25d	refactor(skills): introduce bailian-protocol companion and slim bailian-cli routing	2026-08-04 18:16:30 +08:00
若麒	1e1f5306b3	chore(release): prepare 1.14.0	2026-08-04 18:11:47 +08:00
clh02467605	13158856e8	feat: Refactor skills by granularity and optimize constraints	2026-08-04 15:31:07 +08:00
gujieye	cf2592c07d	Merge pull request #133 from modelstudioai/feat/bailian-wiki-doc-sync feat: add skill commend & wiki sync	2026-08-03 20:07:35 +08:00
故璃	3766b6d7ca	Merge branch 'main' into feat/bailian-wiki-doc-sync	2026-08-03 19:33:16 +08:00
故璃	1962758b0c	feat: add request id	2026-08-03 19:32:27 +08:00
若麒	026e250cd3	Merge branch 'main' into feat/multi-channel-install	2026-08-03 17:16:56 +08:00
Gong Shiqi	6d61afc1d5	Merge pull request #132 from modelstudioai/feat/update-defmodel feat: switch default text model to qwen3.8-max	2026-08-03 16:25:17 +08:00
若麒	7a870ec417	chore(release): prepare 1.13.1	2026-08-03 16:19:00 +08:00
clh02467605	1c38c381e5	feat: switch default text model to qwen3.8-max Align text chat, pipeline, config UI, login validation, and Token Plan text presets, and update README, skill reference, and related tests.	2026-08-03 15:34:18 +08:00
rendianmeng	658763af2c	fix: ci test	2026-08-03 15:29:55 +08:00
rendianmeng	da2ddb7a55	fix: ci test	2026-08-03 14:42:31 +08:00
rendianmeng	be3033baf9	feat: win bl update exe file test	2026-07-31 19:40:53 +08:00
rendianmeng	8ad3e7b947	feat: win bl update exe file test	2026-07-31 19:05:53 +08:00
rendianmeng	525412f566	feat: win bl update exe file test	2026-07-31 18:53:37 +08:00
rendianmeng	75b056ba64	feat: win bl update exe file test	2026-07-31 18:16:53 +08:00
rendianmeng	f5a36b1787	feat: win bl update exe file test	2026-07-31 17:57:28 +08:00
rendianmeng	9fb388b75d	Merge branch 'main' of github.com:modelstudioai/cli into feat/multi-channel-install	2026-07-31 17:47:10 +08:00
rendianmeng	45d468838f	feat: win bl update exe file test	2026-07-31 17:44:59 +08:00
clh02467605	72955d66a7	refactor(skill): update bailian-cli metadata sync to handle multiple skills Enhanced the sync script to update the `metadata.version` for all skills in the `skills` directory, rather than just `bailian-cli`. Improved error handling for missing frontmatter and ensured proper versioning across all skill files.	2026-07-30 15:50:29 +08:00
rendianmeng	389c932390	test(runtime): expect npm --version probe in command pack install Co-authored-by: Cursor <cursoragent@cursor.com>	2026-07-30 11:38:40 +08:00
rendianmeng	6870dc50a6	style: fix AGENTS.md table formatting for vp check Co-authored-by: Cursor <cursoragent@cursor.com>	2026-07-30 10:32:39 +08:00
rendianmeng	54b95ed122	Merge branch main of github.com:modelstudioai/cli into feat/multi-channel-install	2026-07-30 10:20:53 +08:00
rendianmeng	5e2833569a	Merge branch main of github.com:modelstudioai/cli into feat/multi-channel-install	2026-07-30 10:12:42 +08:00
rendianmeng	434aac5b08	docs: install shell md	2026-07-30 10:05:31 +08:00
故璃	e46053b93e	fix(tooling): stop interpolating filenames into staged check Passing staged filenames per-file puts repository paths into the argv of the vp check node process. When an endpoint security agent matches process argv by substring, the whole process is SIGKILLed and pre-commit can never finish. Use the function form so the command runs without filenames: one whole-repo check, wider coverage than per-file, and independent of any path.	2026-07-29 17:54:54 +08:00
故璃	30fe8182f4	Merge branch 'main' into feat/bailian-wiki-doc-sync # Conflicts: # packages/cli/src/commands.ts # packages/commands/tests/e2e/topic-routes.ts # pnpm-lock.yaml # pnpm-workspace.yaml # skills/bailian-cli/reference/index.md	2026-07-29 17:34:28 +08:00
故璃	65c0fe9604	feat: add skill commend & skill install	2026-07-29 17:04:17 +08:00
rendianmeng	fb0c4b81be	docs: install shell md	2026-07-28 14:08:32 +08:00
rendianmeng	952f2277a4	docs: install shell md	2026-07-28 13:51:22 +08:00
rendianmeng	871c667e97	docs: install shell md	2026-07-28 13:49:51 +08:00
clh02467605	4c494207d6	docs(skill): prefer bailian-cli for image/video/audio generation routing Lead the skill description with a dedicated media-generation entry and stronger class-3 priority so agents pick bl for gen/edit tasks, while keeping host-first routing for ordinary text/search.	2026-07-28 10:25:54 +08:00
rendianmeng	af3286dd00	Merge branch 'feat/multi-channel-install' of github.com:modelstudioai/cli into feat/multi-channel-install	2026-07-28 10:20:25 +08:00
rendianmeng	6465c4a78a	feat: install shell test	2026-07-28 10:19:54 +08:00
故璃	467756b319	feat: update manifest.json	2026-07-28 10:17:16 +08:00
故璃	7250de9228	feat: add changelog sync to oss	2026-07-27 16:56:46 +08:00
故璃	51ed69596e	feat: skill update REASON opt	2026-07-27 16:25:49 +08:00
故璃	67b7fa30a7	feat: opt bl skill update commend, keep it atom	2026-07-27 16:02:27 +08:00
故璃	bd17c27023	feat: index.json protocol adapter	2026-07-27 15:41:04 +08:00
故璃	87c37994f2	feat: update skill commend group	2026-07-27 12:30:20 +08:00
故璃	ebbd173b79	feat: update manifest.json path	2026-07-25 08:43:38 +08:00
故璃	6bdc16597b	feat: add secret	2026-07-25 08:07:29 +08:00
故璃	e736bab9c1	feat: add installer sync	2026-07-25 00:33:11 +08:00
故璃	8dd786287f	feat: add skill commend	2026-07-24 19:56:53 +08:00
rendianmeng	d30fb2ae68	feat(release): distribute binaries as per-platform zips	2026-07-24 15:33:15 +08:00
rendianmeng	a1a448c5d2	fix(release): fix binary CI publish and clarify release modules Stabilize Bun compile on 1.2.19, align manifests with OSS consumers, and split gh / webhook / mode helpers out of binary-release.	2026-07-24 10:35:46 +08:00
rendianmeng	7b949d3d3c	fix(release): fix binary CI publish and clarify release modules Stabilize Bun compile on 1.2.19, align manifests with OSS consumers, and split gh / webhook / mode helpers out of binary-release.	2026-07-24 10:34:39 +08:00
rendianmeng	168e2b5ccb	build: multi channel install test	2026-07-23 18:18:46 +08:00
rendianmeng	9fbd2e4ec6	build: multi channel install test	2026-07-23 18:12:42 +08:00
rendianmeng	4bd84e934c	build: multi channel install test	2026-07-23 17:52:52 +08:00
rendianmeng	08bdc3be97	build: multi channel install test	2026-07-23 17:36:14 +08:00
rendianmeng	66a797203c	multi channel install test	2026-07-23 17:34:30 +08:00
故璃	90a44d7140	feat: llm wiki sync	2026-07-23 15:31:57 +08:00
故璃	d08edf0cd8	feat: sync wiki data from oss by fc	2026-07-17 16:43:06 +08:00