Merge pull request #84 from modelstudioai/chore/kscli-version-1.5.0

chore(release): align knowledge-studio-cli to 1.5.0

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -0,0 +1,185 @@
+name: Bug Report
+description: Report a bug in bailian-cli (bl)
+title: "[bug]: "
+labels:
+  - bug
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to report a bug.
+
+        **Before submitting:** search [open issues](https://github.com/modelstudioai/cli/issues?q=is%3Aissue+is%3Aopen) for duplicates.
+
+        **Security:** redact API keys (`sk-...`), console tokens, internal URLs, and business prompts before pasting output.
+
+  - type: markdown
+    attributes:
+      value: |
+        ## Environment
+
+  - type: input
+    id: cli-version
+    attributes:
+      label: CLI version
+      description: "Output of bl --version (use only X.Y.Z, without the bl prefix)"
+      placeholder: "1.2.1"
+    validations:
+      required: true
+
+  - type: input
+    id: skill-version
+    attributes:
+      label: Skill version (optional)
+      description: "metadata.version from the installed bailian-cli skill, if applicable"
+      placeholder: "1.2.1"
+
+  - type: input
+    id: node-version
+    attributes:
+      label: Node version
+      description: "Output of node --version"
+      placeholder: "v22.12.0"
+    validations:
+      required: true
+
+  - type: input
+    id: os
+    attributes:
+      label: OS
+      description: "e.g. darwin 24.5.0, Ubuntu 22.04"
+      placeholder: "darwin 24.5.0"
+    validations:
+      required: true
+
+  - type: dropdown
+    id: region
+    attributes:
+      label: Region
+      description: "From bl auth status or bl config show"
+      options:
+        - cn
+        - us
+        - intl
+        - unknown
+    validations:
+      required: true
+
+  - type: markdown
+    attributes:
+      value: |
+        ## Reproduction
+
+  - type: textarea
+    id: reproduce-command
+    attributes:
+      label: Command to reproduce
+      description: "Exact command that failed. Redact --api-key, sk-..., and sensitive prompts."
+      render: shell
+      placeholder: |
+        bl video generate --prompt "sunset" --download out.mp4 --verbose
+    validations:
+      required: true
+
+  - type: textarea
+    id: expected
+    attributes:
+      label: Expected behavior
+      description: What should have happened?
+    validations:
+      required: true
+
+  - type: textarea
+    id: actual
+    attributes:
+      label: Actual behavior
+      description: What happened instead?
+    validations:
+      required: true
+
+  - type: markdown
+    attributes:
+      value: |
+        ## Error output
+
+        Paste stderr as printed by `bl`. Include `Request ID` when present — it helps us trace logs.
+
+  - type: textarea
+    id: full-output
+    attributes:
+      label: Full output
+      description: Error, Hint, Status, Request ID, Exit code, etc.
+      render: shell
+      placeholder: |
+        Error:       Generation completed but no images returned.
+        Hint:        ...
+        Status:      HTTP 200 (...)
+        Request ID:  ...
+        Exit code:   1
+    validations:
+      required: true
+
+  - type: textarea
+    id: json-error
+    attributes:
+      label: JSON error (optional)
+      description: "Re-run with --output json and paste the error object if available"
+      render: json
+      placeholder: |
+        {
+          "error": {
+            "code": 1,
+            "message": "...",
+            "http_status": 200,
+            "api_code": "...",
+            "request_id": "..."
+          }
+        }
+
+  - type: markdown
+    attributes:
+      value: |
+        ## Troubleshooting already tried
+
+  - type: checkboxes
+    id: already-tried
+    attributes:
+      label: Already tried
+      options:
+        - label: "bl update and skill version aligned with CLI"
+        - label: "bl auth status OK for this command"
+        - label: "Different network / region — still reproduces"
+
+  - type: markdown
+    attributes:
+      value: |
+        ## Additional context
+
+  - type: dropdown
+    id: frequency
+    attributes:
+      label: How often does this happen?
+      options:
+        - Always
+        - Intermittent
+        - Once
+    validations:
+      required: true
+
+  - type: dropdown
+    id: invoked-via
+    attributes:
+      label: How was bl invoked?
+      options:
+        - Terminal (manual)
+        - Agent (Cursor, Claude, etc.)
+        - CI / script
+        - Other
+    validations:
+      required: true
+
+  - type: textarea
+    id: notes
+    attributes:
+      label: Notes (optional)
+      description: Anything else that might help — related issues, screenshots, minimal repro repo, etc.

.github/workflows/ci.yml

@@ -25,6 +25,8 @@ jobs:
 
       - run: pnpm install --frozen-lockfile
 
+      - run: pnpm run sync:skill-assets
+
       - run: pnpm -r --filter "./packages/*" build
 
       - run: pnpm run check

.github/workflows/publish.yml

@@ -3,6 +3,13 @@ name: Publish
 on:
   workflow_dispatch:
     inputs:
+      package:
+        description: "Which package set to publish"
+        required: true
+        type: choice
+        options:
+          - bailian-cli
+          - knowledge-studio-cli
       mode:
         description: "Publish mode"
         required: true
@@ -16,13 +23,13 @@ on:
         type: string
 
 concurrency:
-  group: publish-${{ inputs.mode }}-${{ inputs.channel }}
+  group: publish-${{ inputs.package }}-${{ inputs.mode }}-${{ inputs.channel }}
   cancel-in-progress: false
 
 jobs:
   publish-stable:
     if: inputs.mode == 'stable'
-    name: publish stable to npm + tag
+    name: publish stable (${{ inputs.package }}) to npm + tag
     runs-on: ubuntu-latest
     environment: production # Required Reviewers gate
     permissions:
@@ -51,11 +58,11 @@ jobs:
       - run: pnpm install --frozen-lockfile
 
       - name: publish-stable
-        run: node tools/release/publish-stable.mjs
+        run: node tools/release/publish-stable.mjs ${{ inputs.package == 'knowledge-studio-cli' && '--knowledge' || '' }}
 
   publish-channel:
     if: inputs.mode == 'channel'
-    name: publish beta to npm
+    name: publish channel (${{ inputs.package }}) to npm
     runs-on: ubuntu-latest
     permissions:
       contents: read # no tag, no Release; just publish
@@ -83,4 +90,4 @@ jobs:
       - run: pnpm install --frozen-lockfile
 
       - name: publish-channel
-        run: node tools/release/publish-channel.mjs --channel "${{ inputs.channel }}"
+        run: node tools/release/publish-channel.mjs ${{ inputs.package == 'knowledge-studio-cli' && '--knowledge' || '' }} --channel "${{ inputs.channel }}"

.github/workflows/sync-bailian-cli-skill-reference.yml

@@ -1,143 +0,0 @@
-# When CLI command definitions or the reference generator change, regenerate skill
-# reference markdown, push to modelstudioai/skills, and open a PR against main.
-#
-# Required repository secret (Settings → Secrets and variables → Actions):
-#   SKILLS_SYNC_TOKEN — PAT with repo scope on modelstudioai/skills:
-#     Contents: Read and write
-#     Pull requests: Read and write
-#   Prefer a bot / machine user PAT if your org restricts personal PATs.
-
-name: Sync bailian-cli skill reference
-
-on:
-  push:
-    branches: [main]
-    paths:
-      - "packages/cli/src/commands/**"
-      - "tools/generate-reference.ts"
-      - "packages/core/src/types/command.ts"
-    paths-ignore:
-      # Tests and fixtures under commands/ do not affect generated reference.
-      - "packages/cli/src/commands/**/*.test.ts"
-      - "packages/cli/src/commands/**/*.spec.ts"
-      - "packages/cli/src/commands/**/__fixtures__/**"
-      - "packages/cli/src/commands/**/__tests__/**"
-  schedule:
-    - cron: "0 19 * * *"
-  workflow_dispatch:
-
-concurrency:
-  group: sync-bailian-cli-skill-reference
-  cancel-in-progress: true
-
-jobs:
-  sync:
-    runs-on: ubuntu-latest
-    if: github.repository == 'modelstudioai/cli'
-    permissions:
-      contents: read
-
-    steps:
-      - name: Checkout cli
-        uses: actions/checkout@v4
-
-      - name: Setup pnpm
-        uses: pnpm/action-setup@v4
-        with:
-          version: 10.33.2
-
-      - name: Setup Node
-        uses: actions/setup-node@v4
-        with:
-          node-version: "22"
-          cache: "pnpm"
-
-      - name: Install dependencies
-        run: pnpm install --frozen-lockfile
-
-      - name: Build bailian-cli-core (required by generate-reference)
-        run: pnpm --filter bailian-cli-core run build
-
-      - name: Generate reference markdown
-        run: pnpm --filter bailian-cli run generate:reference
-
-      - name: Checkout skills repo
-        uses: actions/checkout@v4
-        with:
-          repository: modelstudioai/skills
-          path: skills-repo
-          token: ${{ secrets.SKILLS_SYNC_TOKEN }}
-          fetch-depth: 0
-
-      - name: Sync reference into skills and open PR
-        env:
-          GH_TOKEN: ${{ secrets.SKILLS_SYNC_TOKEN }}
-          CLI_SHA: ${{ github.sha }}
-          CLI_RUN: ${{ github.run_id }}
-        run: |
-          set -euo pipefail
-          cd "${GITHUB_WORKSPACE}/skills-repo"
-          git config user.name "github-actions[bot]"
-          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
-
-          git fetch origin main
-          git checkout main
-          git pull origin main
-
-          SHORT_SHA="${CLI_SHA:0:7}"
-          BRANCH="sync/bailian-cli-reference-${SHORT_SHA}"
-          REFERENCE_PATH="skills/bailian-cli/reference"
-
-          git checkout -B "$BRANCH"
-
-          SRC="${GITHUB_WORKSPACE}/tools/generated/reference"
-          DEST="${GITHUB_WORKSPACE}/skills-repo/${REFERENCE_PATH}"
-          mkdir -p "$DEST"
-          rsync -a --delete "$SRC/" "$DEST/"
-
-          # Untracked new files are invisible to `git diff` until added.
-          git add -A -- "$REFERENCE_PATH"
-
-          if git diff --cached --quiet origin/main -- "$REFERENCE_PATH"; then
-            echo "No diff vs origin/main under ${REFERENCE_PATH}; exiting."
-            exit 0
-          fi
-
-          RUN_URL="${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${CLI_RUN}"
-          git commit \
-            -m "chore(bailian-cli): sync reference from cli" \
-            -m "Synced from modelstudioai/cli@${CLI_SHA}" \
-            -m "Workflow run: ${RUN_URL}"
-
-          # Re-fetch before push: main may have been updated with the same reference content.
-          git fetch origin main
-          if git diff --quiet origin/main HEAD -- "$REFERENCE_PATH"; then
-            echo "Committed tree matches origin/main; skipping push and PR."
-            exit 0
-          fi
-
-          git push -u origin "$BRANCH" --force-with-lease
-
-          EXISTING=$(gh pr list --repo modelstudioai/skills --head "$BRANCH" --state open --json number --jq 'length')
-          if [ "${EXISTING}" -eq 0 ]; then
-            BODY_FILE="$(mktemp)"
-            {
-              echo "## Summary"
-              echo ""
-              echo "- Regenerated \`skills/bailian-cli/reference/*.md\` from CLI command definitions at [\`modelstudioai/cli\`](https://github.com/modelstudioai/cli) commit \`${SHORT_SHA}\`."
-              echo ""
-              echo "## Test plan"
-              echo ""
-              echo "- [ ] Spot-check \`reference/index.md\` links and a sample group file under \`skills/bailian-cli/reference/\`."
-              echo "- [ ] Merge if docs only."
-            } >"$BODY_FILE"
-            gh pr create \
-              --repo modelstudioai/skills \
-              --base main \
-              --head "$BRANCH" \
-              --title "chore(bailian-cli): sync CLI command reference" \
-              --body-file "$BODY_FILE"
-            rm -f "$BODY_FILE"
-          else
-            echo "Open PR already exists for head ${BRANCH}."
-          fi

.gitignore

@@ -12,6 +12,7 @@ node_modules
 dist
 dist-ssr
 tools/generated
+.node-version
 
 *.local
 
@@ -33,6 +34,7 @@ tools/generated
 .claude/worktrees/
 .claude/settings.json
 .claude/settings.local.json
+.claude/scheduled_tasks.lock
 .cursor/
 .qwen/
 .playwright-mcp/

.node-version

@@ -0,0 +1 @@
+24

.vite-hooks/pre-commit

@@ -1 +1,10 @@
+#!/usr/bin/env sh
+set -eu
+
+# Regenerate skill reference + SKILL metadata (needs bailian-cli-core dist).
+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
+
 vp staged

AGENTS.md

@@ -25,13 +25,14 @@ packages/cli/
 └── tests/e2e/
 ```
 
-Skill / 命令手册不再随 npm 包发布,改由独立的 `npx add skills` 机制安装。`tools/generate-reference.ts` 仍然从 `catalog.ts` 生成命令手册到 `tools/generated/reference/`(gitignore,临时),等新机制接入后再迁走。
+Skill / 命令手册随 `skills/bailian-cli/` 经 `npx skills add modelstudioai/cli` 安装。`tools/generate-reference.ts` 从 `catalog.ts` 生成命令手册到 `skills/bailian-cli/reference/`(纳入 git);与 `tools/sync-skill-metadata.ts` 一起在 **pre-commit**（`.vite-hooks/pre-commit`）及根脚本 `pnpm run sync:skill-assets` 中执行。
 
 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 `catalog.ts` 生成命令手册(临时输出到 `tools/generated/reference/`)
-- `README.md` / `README_CN.md` — npm 和 GitHub 主页
+- `tools/generate-reference.ts` — 从 `catalog.ts` 生成命令手册到 `skills/bailian-cli/reference/`
+- `tools/sync-skill-metadata.ts` — 从 `packages/cli/package.json` 同步 `skills/bailian-cli/SKILL.md` 的 `metadata.version`（与 `generate:reference` 一并由根目录 `pnpm run sync:skill-assets` 及 pre-commit 执行）
+- `README.md` / `README.zh.md` — npm 和 GitHub 主页
 
 约定:
 
@@ -92,6 +93,18 @@ CLI 只为「自己能权威解释的错误」发出语义化信号,服务端的
 
 不要扮演服务端错误的翻译官——我们没有最新的错误码体系认知,二次包装只会撒谎(详见 `docs/agents/error-hint-change.md` 中的反面 case)。
 
+### 4. Console Gateway 命令必须声明 console 全局 flags
+
+如果新命令使用了 `callConsoleGateway`，必须在 `options` 中添加以下三个全局 flag 的说明，以便 `--help` 中展示：
+
+```ts
+{ flag: "--console-region <region>", description: "Console region" },
+{ flag: "--console-site <site>", description: "Console site: domestic, international" },
+{ flag: "--console-switch-agent <uid>", description: "Switch agent UID", type: "number" },
+```
+
+这些 flag 已在 `GLOBAL_OPTIONS`（`packages/core/src/types/command.ts`）中注册，由 `loadConfig` 写入 `config.consoleRegion` / `config.consoleSite` / `config.consoleSwitchAgent`，`callConsoleGateway` 自动读取——命令无需手动提取或传递。
+
 ## 完成改动后的快速验证
 
 ```sh

CHANGELOG.md

@@ -2,9 +2,145 @@
 
 All notable changes to `bailian-cli` and `bailian-cli-core` are documented here.
 
-The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). The two packages share a single version number — they are always released together.
+The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). The `bailian-cli`, `bailian-cli-core`, `bailian-cli-runtime`, and `bailian-cli-commands` packages share a single version number — they are always released together.
 
-[中文版](CHANGELOG_CN.md) · [README](README.md) · [Contributing](CONTRIBUTING.md)
+[中文版](CHANGELOG.zh.md) · [README](README.md) · [Contributing](CONTRIBUTING.md)
+
+## [1.5.0] - 2026-07-01
+
+### Added
+
+- Model fine-tuning — `bl finetune`: create, list, get, watch, and cancel jobs; fetch training logs; list checkpoints; export a checkpoint as a deployable model; and query training capability (by model or by training type). Supports `sft`, `sft-lora`, `dpo`, `dpo-lora`, and `cpt` training types.
+- Model deployment — `bl deploy`: create, list, get, update (rate limits), scale, and delete deployments; list deployable models and plans.
+- Dataset management — `bl dataset`: upload, list, get, and delete dataset files, plus `bl dataset validate` to check a local `.jsonl` before uploading (ChatML / DPO / CPT formats).
+- Token Plan management — `bl token-plan`: list subscription seats, add members, batch-assign seats, and create a per-seat API key.
+- Automatic update check: after a command finishes, the CLI checks npm for a newer release (throttled) and shows an `Update available` hint; a major stable-version gap upgrades itself automatically. Skipped with `--quiet` or when running `bl update`.
+- Composable packages: `bailian-cli-runtime` (CLI framework) and `bailian-cli-commands` (command library) are now published alongside `bailian-cli-core`, and a new sibling CLI `knowledge-studio-cli` (`kscli`) ships on top of them. `bl` behavior is unchanged.
+
+### Removed
+
+- `bl config export-schema` (exported CLI commands as Anthropic/OpenAI-compatible JSON tool schemas) has been removed.
+
+### Fixed
+
+- Console gateway commands (`bl console call`, etc.) now surface a readable message when the gateway returns a non-string `errorCode`, instead of `[object Object]`.
+
+## [1.4.2] - 2026-06-24
+
+### Added
+
+- `bl omni --list-voices` prints the built-in output voices (ID, name, description, language) and exits without needing an API key. The built-in voice table is expanded from 6 to 17 voices, including dialect voices such as Dylan, Sunny, and Kiki.
+
+### Changed
+
+- `bl omni` default `--voice` is now `Tina` (previously `Cherry`). The `--voice` help points at `--list-voices` instead of listing every option inline.
+- `bl speech synthesize --list-voices` and its missing-`--voice` hint now include a link to the official CosyVoice voice documentation.
+- Agent skill setup guidance now covers console site selection (`--console-site domestic` / `international`) for console login and gateway commands.
+
+### Fixed
+
+- `bl speech synthesize` corrects the `cosyvoice-v3-flash` built-in voice ID from `longanhuan` to `longanhuan_v3`.
+
+## [1.4.1] - 2026-06-22
+
+### Changed
+
+- Video generation now defaults to the upgraded HappyHorse 1.1 model for better quality. The 1.0 models are still available via `--model`.
+- `bl update` now keeps the agent skill in sync across all your agent apps (Claude Code, Cursor, etc.), and refreshes it even when the CLI is already up to date.
+
+## [1.4.0] - 2026-06-17
+
+### Added
+
+- Console gateway now supports multiple regions and sites: `cn-beijing` and `ap-southeast-1`, each with domestic and international variants, plus `switchAgent` for delegated access.
+- New global flags `--console-region`, `--console-site`, and `--console-switch-agent`; `bl console call` also gains `--site` and `--switch-agent`.
+- `bl auth login --base-url <url>` to specify the base URL when logging in with an API key.
+- `bl omni` gains a `--voice` option (Chelsie, Cherry, Ethan, Serena, Sunny, Tina; default Cherry).
+
+### Changed
+
+- All user-facing CLI text is now standardized to English.
+- `bl advisor recommend` internal intent/ranking model upgraded from `qwen-turbo` to `qwen-flash`.
+- Cleaner JSON output for `usage`, `quota`, and `workspace` commands.
+- `base_url` from the config file now takes priority over the `DASHSCOPE_BASE_URL` environment variable.
+- `bl config show` now displays all fields from `config.json`, with sensitive values masked.
+
+### Removed
+
+- The legacy `region` config field and its related options.
+- Invalid leftover code for the removed `model list` command.
+
+### Fixed
+
+- When the console session is not logged in or has expired, the CLI now shows a clear sign-in prompt instead of a generic gateway error.
+- Corrected `--resolution` / `--ratio` / `--duration` flag descriptions for `bl video` commands.
+
+## [1.3.3] - 2026-06-16
+
+### Changed
+
+- `bl knowledge retrieve --help` now clearly indicates that `--api-key` is the recommended authentication method; AK/SK flags are explicitly marked as deprecated with guidance to use `--api-key` instead.
+
+### Added
+
+- `notes` field for command definitions — commands can now include contextual notes (auth requirements, deprecation notices, etc.) that are displayed in both `--help` output and the generated reference docs.
+
+## [1.3.2] - 2026-06-12
+
+### Fixed
+
+- Fixed `bl omni --audio` always returning HTTP 400 (#54); audio inputs are now understood correctly.
+
+## [1.3.1] - 2026-06-12
+
+### Fixed
+
+- `bl` now honors `HTTP_PROXY` / `HTTPS_PROXY` / `NO_PROXY` environment variables (#35). Node's built-in `fetch` (undici) ignores proxy env vars by default, causing `ECONNRESET` for users behind a VPN or corporate proxy. A global proxy dispatcher is now installed at startup when these variables are set, and the `ECONNRESET` error hint points to `export HTTPS_PROXY=http://127.0.0.1:<port>`.
+
+## [1.3.0] - 2026-06-10
+
+### Added
+
+- `bl knowledge retrieve` now supports API-Key authentication (DashScope gateway), in addition to AK/SK. API-Key is auto-detected and preferred when available.
+- New retrieval options: `--dense-similarity-top-k`, `--sparse-similarity-top-k`, `--rerank-model`, `--rerank-mode`, `--rerank-instruct` — supported on both API-Key and AK/SK paths.
+- `DashScopeKnowledgeRetrieveRequest` / `DashScopeKnowledgeRetrieveResponse` types and `knowledgeRetrieveEndpoint` added to `bailian-cli-core`.
+- Comprehensive E2E tests for knowledge retrieve covering both auth paths, dry-run, rerank flags, and error cases.
+
+- `bl usage` command group:
+  - `bl usage free` — query free-tier quota for all models (or a specific model with `--model`).
+  - `bl usage freetier` — enable (`--on`) or disable (`--off`) auto-stop for free-tier models.
+  - `bl usage stats` — query model usage statistics (requires `--workspace-id`).
+- `bl quota` command group:
+  - `bl quota list` — view model RPM/TPM rate limits (filter with `--model`, show all with `--all`).
+  - `bl quota check` — check current RPM/TPM usage against rate limits.
+  - `bl quota history` — view quota change history with pagination.
+  - `bl quota request` — request a temporary quota increase for a model.
+- `bl workspace list` — list all workspaces with region and endpoint details.
+
+### Changed
+
+- Credential resolution priority: explicit API-Key → explicit AK/SK flags → auto-detected API-Key → fallback AK/SK from config/env.
+- `--workspace-id` is now only required for AK/SK auth, no longer mandatory for API-Key mode.
+- `--top-k` deprecated in favor of `--rerank-top-n`; emits a warning and maps to `--rerank-top-n` when used.
+- `--access-key-id` / `--access-key-secret` flags marked as deprecated (API-Key is recommended).
+- API Key and console links updated to direct key management pages across all docs.
+
+### Fixed
+
+- `--rerank` flag in AK/SK path now correctly sets `EnableReranking` instead of the non-functional `Rerank: true` boolean.
+
+## [1.2.1] - 2026-06-09
+
+### Changed
+
+- Skill install command updated from `npx skills add modelstudioai/skills` to `npx skills add modelstudioai/cli --all -g` across all READMEs and docs.
+- `bl update` now automatically updates the `bailian-cli` agent skill after CLI upgrade.
+- Renamed `README_CN.md` to `README.zh.md` (ISO 639 convention) across the entire repo.
+
+### Added
+
+- Official skill (`skills/bailian-cli/`) now ships in this repository with pre-commit auto-generation of reference docs and SKILL.md version sync.
+- Bilingual READMEs (EN + CN) for the `bailian-cli` skill.
 
 ## [1.2.0] - 2026-06-05
 

CHANGELOG.zh.md

@@ -0,0 +1,250 @@
+# 更新日志
+
+`bailian-cli` 和 `bailian-cli-core` 的所有重要变更都记录在此。
+
+格式遵循 [Keep a Changelog](https://keepachangelog.com/zh-CN/1.1.0/),版本号遵循 [语义化版本](https://semver.org/lang/zh-CN/spec/v2.0.0.html)。`bailian-cli`、`bailian-cli-core`、`bailian-cli-runtime`、`bailian-cli-commands` 共享一个版本号,总是一起发布。
+
+[English](CHANGELOG.md) · [README](README.zh.md) · [参与贡献](CONTRIBUTING.zh.md)
+
+## [1.5.0] - 2026-07-01
+
+### 新增
+
+- 模型精调 —— `bl finetune`:创建、列出、查询、观察、取消训练任务;拉取训练日志;列出 checkpoint;将 checkpoint 导出为可部署模型;查询训练能力(按模型或按训练类型)。支持 `sft`、`sft-lora`、`dpo`、`dpo-lora`、`cpt` 训练类型。
+- 模型部署 —— `bl deploy`:创建、列出、查询、更新(限流)、扩缩容、删除部署;列出可部署模型与套餐。
+- 数据集管理 —— `bl dataset`:上传、列出、查询、删除数据集文件,并新增 `bl dataset validate` 在上传前本地校验 `.jsonl`(ChatML / DPO / CPT 格式)。
+- Token Plan 管理 —— `bl token-plan`:列出订阅座位、添加成员、批量分配座位、为座位创建 API Key。
+- 自动更新检查:命令执行完成后,CLI 会(节流地)检查 npm 上是否有新版本并提示 `Update available`;若与稳定版存在大版本差距则自动升级。`--quiet` 或执行 `bl update` 时跳过。
+- 可组合包:`bailian-cli-runtime`(CLI 框架)与 `bailian-cli-commands`(命令库)现在与 `bailian-cli-core` 一起发布,并在其之上新增了同家族 CLI `knowledge-studio-cli`(`kscli`)。`bl` 行为保持不变。
+
+### 已移除
+
+- 移除 `bl config export-schema` 命令(原用于把 CLI 命令导出为 Anthropic/OpenAI 兼容的 JSON tool schema)。
+
+### 修复
+
+- 控制台网关类命令(`bl console call` 等)在网关返回非字符串 `errorCode` 时,现在会给出可读的错误信息,而不是 `[object Object]`。
+
+## [1.4.2] - 2026-06-24
+
+### 新增
+
+- `bl omni --list-voices` 无需 API key 即可打印内置输出音色列表(ID、名称、描述、语言)并退出。内置音色表从 6 个扩展到 17 个,新增 Dylan、Sunny、Kiki 等方言音色。
+
+### 变更
+
+- `bl omni` 默认 `--voice` 改为 `Tina`(原为 `Cherry`)。`--voice` 帮助文案改为指向 `--list-voices`,不再内联列出全部音色。
+- `bl speech synthesize --list-voices` 输出及缺少 `--voice` 时的提示中,新增官方 CosyVoice 音色文档链接。
+- Agent skill 配置指引新增 console 站点选择说明(`--console-site domestic` / `international`),适用于 console 登录与网关类命令。
+
+### 修复
+
+- `bl speech synthesize` 修正 `cosyvoice-v3-flash` 内置音色 ID,由 `longanhuan` 改为 `longanhuan_v3`。
+
+## [1.4.1] - 2026-06-22
+
+### 变更
+
+- 视频生成默认升级到 HappyHorse 1.1 模型,画面质量更佳。如需使用 1.0 模型,可通过 `--model` 指定。
+- `bl update` 现在会把 agent skill 同步更新到所有 agent 应用(Claude Code、Cursor 等),即使 CLI 已是最新版本也会刷新 skill。
+
+## [1.4.0] - 2026-06-17
+
+### 新增
+
+- 控制台网关支持多 region 与多站点：`cn-beijing` 与 `ap-southeast-1`，各含国内站 / 国际站变体，并新增 `switchAgent` 委托访问。
+- 新增全局标志 `--console-region`、`--console-site`、`--console-switch-agent`；`bl console call` 另外新增 `--site` 与 `--switch-agent`。
+- `bl auth login --base-url <url>`：使用 API Key 登录时可指定 base URL。
+- `bl omni` 新增 `--voice` 选项（Chelsie、Cherry、Ethan、Serena、Sunny、Tina，默认 Cherry）。
+
+### 变更
+
+- 所有面向用户的 CLI 文案统一为英文。
+- `bl advisor recommend` 内部意图 / 排序模型由 `qwen-turbo` 升级为 `qwen-flash`。
+- 优化 `usage`、`quota`、`workspace` 命令的 JSON 输出。
+- 配置文件中的 `base_url` 现在优先级高于环境变量 `DASHSCOPE_BASE_URL`。
+- `bl config show` 现在展示 `config.json` 中的全部字段（敏感值已脱敏）。
+
+### 移除
+
+- 移除遗留的 `region` 配置字段及其相关选项。
+- 清理 `model list` 命令移除后遗留的无效代码。
+
+### 修复
+
+- 当控制台会话未登录或已过期时，CLI 现在会给出明确的登录提示，不再是笼统的网关错误。
+- 修正 `bl video` 命令 `--resolution` / `--ratio` / `--duration` 的帮助文案。
+
+## [1.3.3] - 2026-06-16
+
+### 变更
+
+- `bl knowledge retrieve --help` 现在明确指出 `--api-key` 是推荐的鉴权方式；AK/SK 相关选项已标注废弃并引导用户使用 `--api-key`。
+
+### 新增
+
+- 命令定义新增 `notes` 字段 — 命令可以附带上下文说明（鉴权要求、废弃提示等），同时展示在 `--help` 输出和生成的命令手册中。
+
+## [1.3.2] - 2026-06-12
+
+### 修复
+
+- 修复 `bl omni --audio` 始终返回 HTTP 400 的问题（#54），音频输入现已能正常理解。
+
+## [1.3.1] - 2026-06-12
+
+### 修复
+
+- `bl` 现在会读取 `HTTP_PROXY` / `HTTPS_PROXY` / `NO_PROXY` 环境变量(#35)。Node 内置的 `fetch`(undici)默认忽略代理环境变量,导致 VPN 或公司代理下出现 `ECONNRESET`。现已在启动时根据这些变量安装全局代理 dispatcher,并在 `ECONNRESET` 报错提示中给出 `export HTTPS_PROXY=http://127.0.0.1:<port>` 的指引。
+
+## [1.3.0] - 2026-06-11
+
+### 新增
+
+- `bl usage` 命令组：
+  - `bl usage free` — 查询所有模型的免费额度（可通过 `--model` 指定模型）。
+  - `bl usage freetier` — 启用（`--on`）或禁用（`--off`）免费额度模型的自动停服。
+  - `bl usage stats` — 查询模型用量统计（需指定 `--workspace-id`）。
+- `bl quota` 命令组：
+  - `bl quota list` — 查看模型 RPM/TPM 速率限制（支持 `--model` 过滤，`--all` 展示全部）。
+  - `bl quota check` — 查看当前 RPM/TPM 用量与速率限制。
+  - `bl quota history` — 查看配额变更记录，支持分页。
+  - `bl quota request` — 申请模型临时配额提升。
+- `bl workspace list` — 列出所有业务空间，包含地域和 endpoint 信息。
+- `bl knowledge retrieve` 新增 API-Key 鉴权（DashScope 网关），与原有 AK/SK 并存，可用时自动优先使用 API-Key。
+- 新增检索参数：`--dense-similarity-top-k`、`--sparse-similarity-top-k`、`--rerank-model`、`--rerank-mode`、`--rerank-instruct`，API-Key 与 AK/SK 两条链路均支持。
+- `bailian-cli-core` 新增 `DashScopeKnowledgeRetrieveRequest` / `DashScopeKnowledgeRetrieveResponse` 类型及 `knowledgeRetrieveEndpoint` 端点。
+- 知识库检索全面 E2E 测试，覆盖两种鉴权路径、dry-run、rerank 参数及错误场景。
+
+### 变更
+
+- 凭据解析优先级：显式 API-Key → 显式 AK/SK flag → 自动检测 API-Key → 回退至配置/环境变量中的 AK/SK。
+- `--workspace-id` 仅在 AK/SK 鉴权时必填，API-Key 模式下不再强制要求。
+- `--top-k` 标记为废弃，改用 `--rerank-top-n`；使用时输出警告并自动映射。
+- `--access-key-id` / `--access-key-secret` 标记为废弃（推荐使用 API-Key）。
+- 全部文档中的 API Key 和控制台链接更新为直达密钥管理页面。
+
+### 修复
+
+- AK/SK 链路 `--rerank` 现在正确设置 `EnableReranking`，而非之前无效的 `Rerank: true` 布尔值。
+
+## [1.2.1] - 2026-06-09
+
+### 变更
+
+- Skill 安装命令从 `npx skills add modelstudioai/skills` 更新为 `npx skills add modelstudioai/cli --all -g`，所有 README 和文档已同步。
+- `bl update` 现在会在 CLI 升级后自动更新 `bailian-cli` agent skill。
+- 全仓库 `README_CN.md` 统一重命名为 `README.zh.md`（ISO 639 命名规范）。
+
+### 新增
+
+- 官方 skill（`skills/bailian-cli/`）迁入本仓库，pre-commit 自动生成 reference 文档并同步 SKILL.md 版本号。
+- `bailian-cli` skill 新增中英文双语 README。
+
+## [1.2.0] - 2026-06-05
+
+### 新增
+
+- `bl mcp` 命令组：`bl mcp list` 列出 MCP 服务器，`bl mcp tools <server>` 查看可用工具，`bl mcp call <server>.<tool>` 通过 `--arg k=v` 或 `--json` 调用工具。
+- `bl advisor recommend` — 用自然语言描述任务需求，智能推荐最合适的模型，展示上下文窗口、定价及能力详情。
+
+### 修复
+
+- 图片/视频水印始终开启的问题，现在正确遵守 `bl config set watermark false` 配置。
+- 成对 flag（如 `--watermark` / `--no-watermark`）现已正确互斥。
+- 可选参数为空时 flag 校验不再崩溃。
+- **安全**：凭据不再泄漏到磁盘日志，文件权限已收紧。
+- **安全**：校验 `base_url` / `console_gateway_url` 为合法 HTTP(S) URL。
+- **安全**：script/JS `code` 字段强制为字符串字面量（阻止不可信代码 RCE）。
+- **安全**：URL 路径段已百分号编码，SSE 缓冲区设上限。
+- **安全**：流水线规划、指针遍历及并发安全加固。
+- MCP 命令现在在参数校验和 dry-run 检查之后才处理鉴权。
+
+### 变更
+
+- 所有命令的 flag 默认值文案统一并去重。
+- 非法/未知 flag 名称现在会报明确错误，而非静默忽略。
+
+## [1.1.3] - 2026-06-02
+
+### 新增
+
+- `bl auth login --console` 在未配置 DashScope API Key 时会自动获取并保存,一次浏览器登录即可完成 OAuth 与 API Key 配置。
+
+### 变更
+
+- API Key 校验更稳健:网络 / 401 / 5xx 等瞬时错误会自动重试,单次请求超时上限收紧为 30 秒。
+
+## [1.1.2] - 2026-05-29
+
+### 变更
+
+- 默认视觉模型由 `qwen-vl` 升级为 `qwen3-vl-plus`,视觉推理与图表/文档解析能力更强。
+
+### 修复
+
+- 修复 1.1.0 开源切换后暴露的 TypeScript / lint 问题。
+
+## [1.1.1] - 2026-05-29
+
+仅文档更新，CLI 与 SDK 行为无变化。
+
+### 新增
+
+- 新增 `INSTALL.md`，提供面向 AI Agent 的安装指引。
+
+### 变更
+
+- 同步根目录与 `packages/cli` 的 README 互链；中文 README 与英文版对齐。
+- 移除 README 中的 unpkg 链接,改用官方来源。
+- `tools/release.mjs` 在发布前会校验根目录与 `packages/cli` 的 README 保持同步。
+
+### 修复
+
+- `tools/release.mjs check` 现在会先构建包再执行类型检查,确保 `bailian-cli-core` 在干净检出环境下能正确解析(此前会级联出约 80 个虚假的 TS 错误)。
+
+## [1.1.0] - 2026-05-28
+
+GitHub 上的首次公开发布。本项目此前在内部开发,这是首个以 Apache-2.0 协议开源的版本。
+
+### 新增
+
+让您的 AI Agent 开箱就具备以下能力,并可在复杂任务中自动组合调用:
+
+**模型服务**
+
+| 能力       | 默认服务                    | 简介                                                             |
+| ---------- | --------------------------- | ---------------------------------------------------------------- |
+| 文本生成   | `qwen3.7-max`               | 面向智能体时代的旗舰 Max 模型,编程、办公与长周期自主执行能力出色 |
+| 语音生成   | `cosyvoice-v3-flash`        | 多音色实时流式合成,自然度/情感增强,5-20s 样本即可克隆            |
+| 语音识别   | `fun-asr`                   | 汉语七大方言 + 20+ 口音官话,覆盖 30 种语种                       |
+| 图像生成   | `qwen-image-2.0`            | 图片生成与编辑融合,专业文字渲染、真实质感、强语义遵循            |
+| 图像编辑   | `qwen-image-2.0`            | 智能编辑,支持多图合成                                            |
+| 图生视频   | `happyhorse-1.0-i2v`        | 精准理解文本语义,输出流畅自然的高质量视频                        |
+| 文生视频   | `happyhorse-1.0-t2v`        | 高度还原动态画面,细节丰富                                        |
+| 参考生视频 | `happyhorse-1.0-r2v`        | 支持最多 9 张图片参考,稳定主体与场景保持                         |
+| 视频编辑   | `happyhorse-1.0-video-edit` | 自然语言指令编辑视频,支持最多 5 张图片参考                       |
+| 视觉理解   | `qwen-vl`                   | 长视频分析、图表/文档解析、视觉推理、多语言 OCR                  |
+
+**应用数据**
+
+| 能力   | 默认服务         | 简介                                           |
+| ------ | ---------------- | ---------------------------------------------- |
+| 知识库 | 阿里云百炼知识库 | 多模态数据知识库增删改查检索,需 AccessKey 认证 |
+| 记忆库 | 阿里云百炼记忆库 | 跨会话持久化存储,提供个性化连贯对话体验        |
+
+**应用构建**
+
+| 能力       | 默认服务   | 简介                     |
+| ---------- | ---------- | ------------------------ |
+| 工作流调用 | 工作流服务 | 调用已有的工作流应用服务 |
+| 智能体调用 | 智能体服务 | 调用已有的智能体应用服务 |
+
+**工具能力**
+
+| 能力         | 默认服务                            | 简介                                                              |
+| ------------ | ----------------------------------- | ----------------------------------------------------------------- |
+| 联网搜索     | `bailian_web_search`                | 实时互联网全栈信息检索,提升回答准确性及时效性                     |
+| 临时文件上传 | 临时文件上传服务                    | 免费临时存储空间,上传本地文件获得 URL(有效期 48 小时)             |
+| 模型额度查询 | 模型额度查询                        | 根据模型 id 查询可以使用的免费额度                                |
+| 接口文档     | 阿里云百炼模型应用 API 调用参考文档 | 在构建应用的过程中,自动为您的应用集成阿里云百炼模型和应用能力 API |

CHANGELOG_CN.md

@@ -1,115 +0,0 @@
-# 更新日志
-
-`bailian-cli` 和 `bailian-cli-core` 的所有重要变更都记录在此。
-
-格式遵循 [Keep a Changelog](https://keepachangelog.com/zh-CN/1.1.0/),版本号遵循 [语义化版本](https://semver.org/lang/zh-CN/spec/v2.0.0.html)。两个包共享一个版本号,总是一起发布。
-
-[English](CHANGELOG.md) · [README](README_CN.md) · [参与贡献](CONTRIBUTING_CN.md)
-
-## [1.2.0] - 2026-06-05
-
-### 新增
-
-- `bl mcp` 命令组：`bl mcp list` 列出 MCP 服务器，`bl mcp tools <server>` 查看可用工具，`bl mcp call <server>.<tool>` 通过 `--arg k=v` 或 `--json` 调用工具。
-- `bl advisor recommend` — 用自然语言描述任务需求，智能推荐最合适的模型，展示上下文窗口、定价及能力详情。
-
-### 修复
-
-- 图片/视频水印始终开启的问题，现在正确遵守 `bl config set watermark false` 配置。
-- 成对 flag（如 `--watermark` / `--no-watermark`）现已正确互斥。
-- 可选参数为空时 flag 校验不再崩溃。
-- **安全**：凭据不再泄漏到磁盘日志，文件权限已收紧。
-- **安全**：校验 `base_url` / `console_gateway_url` 为合法 HTTP(S) URL。
-- **安全**：script/JS `code` 字段强制为字符串字面量（阻止不可信代码 RCE）。
-- **安全**：URL 路径段已百分号编码，SSE 缓冲区设上限。
-- **安全**：流水线规划、指针遍历及并发安全加固。
-- MCP 命令现在在参数校验和 dry-run 检查之后才处理鉴权。
-
-### 变更
-
-- 所有命令的 flag 默认值文案统一并去重。
-- 非法/未知 flag 名称现在会报明确错误，而非静默忽略。
-
-## [1.1.3] - 2026-06-02
-
-### 新增
-
-- `bl auth login --console` 在未配置 DashScope API Key 时会自动获取并保存,一次浏览器登录即可完成 OAuth 与 API Key 配置。
-
-### 变更
-
-- API Key 校验更稳健:网络 / 401 / 5xx 等瞬时错误会自动重试,单次请求超时上限收紧为 30 秒。
-
-## [1.1.2] - 2026-05-29
-
-### 变更
-
-- 默认视觉模型由 `qwen-vl` 升级为 `qwen3-vl-plus`,视觉推理与图表/文档解析能力更强。
-
-### 修复
-
-- 修复 1.1.0 开源切换后暴露的 TypeScript / lint 问题。
-
-## [1.1.1] - 2026-05-29
-
-仅文档更新，CLI 与 SDK 行为无变化。
-
-### 新增
-
-- 新增 `INSTALL.md`，提供面向 AI Agent 的安装指引。
-
-### 变更
-
-- 同步根目录与 `packages/cli` 的 README 互链；中文 README 与英文版对齐。
-- 移除 README 中的 unpkg 链接,改用官方来源。
-- `tools/release.mjs` 在发布前会校验根目录与 `packages/cli` 的 README 保持同步。
-
-### 修复
-
-- `tools/release.mjs check` 现在会先构建包再执行类型检查,确保 `bailian-cli-core` 在干净检出环境下能正确解析(此前会级联出约 80 个虚假的 TS 错误)。
-
-## [1.1.0] - 2026-05-28
-
-GitHub 上的首次公开发布。本项目此前在内部开发,这是首个以 Apache-2.0 协议开源的版本。
-
-### 新增
-
-让您的 AI Agent 开箱就具备以下能力,并可在复杂任务中自动组合调用:
-
-**模型服务**
-
-| 能力       | 默认服务                    | 简介                                                             |
-| ---------- | --------------------------- | ---------------------------------------------------------------- |
-| 文本生成   | `qwen3.7-max`               | 面向智能体时代的旗舰 Max 模型,编程、办公与长周期自主执行能力出色 |
-| 语音生成   | `cosyvoice-v3-flash`        | 多音色实时流式合成,自然度/情感增强,5-20s 样本即可克隆            |
-| 语音识别   | `fun-asr`                   | 汉语七大方言 + 20+ 口音官话,覆盖 30 种语种                       |
-| 图像生成   | `qwen-image-2.0`            | 图片生成与编辑融合,专业文字渲染、真实质感、强语义遵循            |
-| 图像编辑   | `qwen-image-2.0`            | 智能编辑,支持多图合成                                            |
-| 图生视频   | `happyhorse-1.0-i2v`        | 精准理解文本语义,输出流畅自然的高质量视频                        |
-| 文生视频   | `happyhorse-1.0-t2v`        | 高度还原动态画面,细节丰富                                        |
-| 参考生视频 | `happyhorse-1.0-r2v`        | 支持最多 9 张图片参考,稳定主体与场景保持                         |
-| 视频编辑   | `happyhorse-1.0-video-edit` | 自然语言指令编辑视频,支持最多 5 张图片参考                       |
-| 视觉理解   | `qwen-vl`                   | 长视频分析、图表/文档解析、视觉推理、多语言 OCR                  |
-
-**应用数据**
-
-| 能力   | 默认服务         | 简介                                           |
-| ------ | ---------------- | ---------------------------------------------- |
-| 知识库 | 阿里云百炼知识库 | 多模态数据知识库增删改查检索,需 AccessKey 认证 |
-| 记忆库 | 阿里云百炼记忆库 | 跨会话持久化存储,提供个性化连贯对话体验        |
-
-**应用构建**
-
-| 能力       | 默认服务   | 简介                     |
-| ---------- | ---------- | ------------------------ |
-| 工作流调用 | 工作流服务 | 调用已有的工作流应用服务 |
-| 智能体调用 | 智能体服务 | 调用已有的智能体应用服务 |
-
-**工具能力**
-
-| 能力         | 默认服务                            | 简介                                                              |
-| ------------ | ----------------------------------- | ----------------------------------------------------------------- |
-| 联网搜索     | `bailian_web_search`                | 实时互联网全栈信息检索,提升回答准确性及时效性                     |
-| 临时文件上传 | 临时文件上传服务                    | 免费临时存储空间,上传本地文件获得 URL(有效期 48 小时)             |
-| 模型额度查询 | 模型额度查询                        | 根据模型 id 查询可以使用的免费额度                                |
-| 接口文档     | 阿里云百炼模型应用 API 调用参考文档 | 在构建应用的过程中,自动为您的应用集成阿里云百炼模型和应用能力 API |

CONTRIBUTING.md

@@ -2,7 +2,7 @@
 
 Developer guide for `bailian-cli` — the official CLI for Aliyun Model Studio (DashScope). For end-user usage, see [README.md](README.md).
 
-[中文版](CONTRIBUTING_CN.md) · [README](README.md) · [Changelog](CHANGELOG.md)
+[中文版](CONTRIBUTING.zh.md) · [README](README.md) · [Changelog](CHANGELOG.md)
 
 ## Prerequisites
 

CONTRIBUTING.zh.md

@@ -1,8 +1,8 @@
 # 参与贡献 bailian-cli
 
-`bailian-cli` 是阿里云百炼(DashScope)的官方 CLI。本文是面向**开发者**的指南;终端用户请看 [README_CN.md](README_CN.md)。
+`bailian-cli` 是阿里云百炼(DashScope)的官方 CLI。本文是面向**开发者**的指南;终端用户请看 [README.zh.md](README.zh.md)。
 
-[English](CONTRIBUTING.md) · [README](README_CN.md) · [更新日志](CHANGELOG_CN.md)
+[English](CONTRIBUTING.md) · [README](README.zh.md) · [更新日志](CHANGELOG.zh.md)
 
 ## 环境要求
 

INSTALL.md

@@ -50,10 +50,10 @@ which bl   # Windows 可用 where bl
 CLI 校验通过后，在本机终端执行：
 
 ```bash
-npx skills add modelstudioai/skills --all -g
+npx skills add modelstudioai/cli --all -g
 ```
 
-完整说明见 [百炼Skills](https://github.com/modelstudioai/skills)。安装成功后，用中文简要说明已安装的 skills 及用户可做什么。
+安装成功后，用中文简要说明已安装的 skills 及用户可做什么。
 
 ---
 
@@ -102,7 +102,7 @@ bl auth status --output json
 bl text chat --message "ping" --non-interactive --output json
 ```
 
-若失败：根据 stderr / JSON 中的 `hint` 或 `message` 排查（网络、Key 无效、region 等）。全局 region：`--region cn|us|intl`，默认 `cn`。
+若失败：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`。
 
 ---
 

README.md

@@ -9,7 +9,7 @@
 [![TypeScript](https://img.shields.io/badge/TypeScript-strict-3178c6)](https://www.typescriptlang.org)
 [![License](https://img.shields.io/badge/license-Apache%202.0-blue)](LICENSE)
 
-[Aliyun Model Studio CLI Site](https://bailian.console.aliyun.com/cli?source_channel=cli_github&) · [中文文档](https://github.com/modelstudioai/cli/blob/main/README_CN.md) · [API Documentation](https://help.aliyun.com/zh/model-studio/) · [Get API Key](https://bailian.console.aliyun.com/cli?source_channel=key_github&)
+[Aliyun Model Studio CLI Site](https://bailian.console.aliyun.com/cli?source_channel=cli_github&) · [中文文档](https://github.com/modelstudioai/cli/blob/main/README.zh.md) · [API Documentation](https://help.aliyun.com/zh/model-studio/) · [Get API Key](https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key)
 
 ---
 
@@ -27,15 +27,19 @@ Equip your AI Agent out-of-the-box with these capabilities, composable across co
 - **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-1.0 series: text-/image-/reference-to-video and natural-language video editing (up to 9-image reference)
+- **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
+
+> **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
-- **Console capabilities** — Browse Bailian apps (`app list`) and check free-tier quota (`usage free`)
+- **Fine-tuning & deployment** — Upload datasets, create SFT/LoRA/DPO/CPT jobs (`finetune create`), probe job status non-blockingly (`finetune watch`), query per-model training capability (`finetune capability`), and deploy trained models as endpoints (`deploy create`)
+- **Console capabilities** — Browse Bailian apps (`app list`), 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
@@ -51,7 +55,7 @@ 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 1.0**, Aliyun Model Studio's text-/image-/reference-to-video generation model
+- **[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
 - **[spark-video Skill](https://github.com/JohnKeating1997/spark-video)** — handles scene decomposition, storyboarding, shot continuity, and final stitching
 
 ### The single prompt
@@ -64,7 +68,7 @@ A complete **2-minute, 16:9 cinematic short film** — produced end-to-end from
 
 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 1.0** in parallel.
+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.
 
 No timeline scrubbing. No frame-by-frame editing. Just one sentence → one video.
@@ -73,7 +77,7 @@ No timeline scrubbing. No frame-by-frame editing. Just one sentence → one vide
 
 ```bash
 npm install -g bailian-cli
-npx skills add modelstudioai/skills --all -g
+npx skills add modelstudioai/cli --all -g
 ```
 
 > Requires Node.js >= 22.12.
@@ -108,9 +112,30 @@ bl advisor recommend --message "qwen-max vs deepseek-v3 for code generation"
 # Browser login (required for console capability commands)
 bl auth login --console
 
-# Browse apps / free-tier quota
+# Fine-tune & deploy — a one-shot train-to-serve workflow
+bl dataset upload --file ./train.jsonl                 # Upload a .jsonl dataset (validated first)
+bl finetune 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 status probe (exit 0/1/3 = done/failed/running)
+bl finetune capability --model qwen3-8b               # Which training types a model supports
+bl deploy create --model qwen3-8b --name my-svc --plan mu  # Deploy the trained model as an endpoint
+
+# Browse apps / free-tier quota / usage statistics / workspaces
 bl app list
-bl usage free --model qwen3-max
+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
 ```
 
 > More examples and scenarios: [Aliyun Model Studio CLI Site](https://bailian.console.aliyun.com/cli?source_channel=cli_github&)
@@ -119,7 +144,7 @@ bl usage free --model qwen3-max
 
 ### DashScope API Key
 
-Required for most commands. Get your key from the [DashScope Console](https://bailian.console.aliyun.com/cli?source_channel=key_github&).
+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
@@ -134,15 +159,15 @@ bl text chat --api-key sk-xxxxx --message "Hello"
 
 ### Console Login (OAuth)
 
-Required for console capability commands (`app list`, `usage free`). Opens the Bailian console in your browser to sign in.
+Required for console capability commands (`app list`, `usage free`, `usage stats`, `workspace list`, `quota list/request/check/history`). Opens the Bailian console in your browser to sign in.
 
 ```bash
 bl auth login --console
 ```
 
-### Alibaba Cloud AK/SK (Knowledge Base only)
+### Alibaba Cloud AK/SK (Knowledge Base & Token Plan)
 
-Required for `knowledge retrieve`. Get your AccessKey from [RAM Console](https://ram.console.aliyun.com/manage/ak).
+Required for `knowledge retrieve` and the `token-plan` command group. Get your AccessKey from [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.
 
@@ -159,7 +184,7 @@ export BAILIAN_WORKSPACE_ID=ws-...
 bl config show
 
 # Set defaults
-bl config set --key region --value us
+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
 
@@ -171,14 +196,14 @@ Config file location: `~/.bailian/config.json`
 
 ## Links
 
-| Resource                     | URL                                                               |
-| :--------------------------- | :---------------------------------------------------------------- |
-| Aliyun Model Studio CLI Site | https://bailian.console.aliyun.com/cli?source_channel=cli_github& |
-| DashScope API Docs           | https://help.aliyun.com/zh/model-studio/                          |
-| Qwen Model List              | https://help.aliyun.com/zh/model-studio/getting-started/models    |
-| Aliyun Model Studio Console  | https://bailian.console.aliyun.com/                               |
-| Get API Key                  | https://bailian.console.aliyun.com/cli?source_channel=key_github& |
-| Get AccessKey                | https://ram.console.aliyun.com/manage/ak                          |
+| Resource                     | URL                                                                                       |
+| :--------------------------- | :---------------------------------------------------------------------------------------- |
+| Aliyun Model Studio CLI Site | https://bailian.console.aliyun.com/cli?source_channel=cli_github&                         |
+| DashScope API Docs           | https://help.aliyun.com/zh/model-studio/                                                  |
+| Qwen Model List              | https://help.aliyun.com/zh/model-studio/getting-started/models                            |
+| Aliyun Model Studio Console  | https://bailian.console.aliyun.com/?source_channel=cli_github                             |
+| Get API Key                  | https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key |
+| Get AccessKey                | https://ram.console.aliyun.com/manage/ak                                                  |
 
 ## Changelog
 

README.zh.md

@@ -9,7 +9,7 @@
 [![TypeScript](https://img.shields.io/badge/TypeScript-strict-3178c6)](https://www.typescriptlang.org)
 [![License](https://img.shields.io/badge/license-Apache%202.0-blue)](LICENSE)
 
-[阿里云百炼 CLI 官方主页](https://bailian.console.aliyun.com/cli?source_channel=cli_github&) · [English](https://github.com/modelstudioai/cli/blob/main/README.md) · [API 文档](https://help.aliyun.com/zh/model-studio/) · [获取 API Key](https://bailian.console.aliyun.com/cli?source_channel=key_github&)
+[阿里云百炼 CLI 官方主页](https://bailian.console.aliyun.com/cli?source_channel=cli_github&) · [English](https://github.com/modelstudioai/cli/blob/main/README.md) · [API 文档](https://help.aliyun.com/zh/model-studio/) · [获取 API Key](https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key)
 
 ---
 
@@ -27,15 +27,19 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 - **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-1.0 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
+- **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
+
+> **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
-- **控制台能力** — 浏览百炼应用（`app list`），查询模型免费额度（`usage free`）
+- **微调与部署** — 上传数据集、创建 SFT/LoRA/DPO/CPT 调优任务（`finetune create`）、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 create`）
+- **控制台能力** — 浏览百炼应用（`app list`），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
@@ -51,7 +55,7 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 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 1.0**,百炼的文生/图生/参考生视频模型
+- **[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
@@ -62,7 +66,7 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 
 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 1.0**。
+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。
 
 没有时间线拖拽,没有逐帧剪辑。一句话 → 一部短片。
@@ -71,7 +75,7 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 
 ```bash
 npm install -g bailian-cli
-npx skills add modelstudioai/skills --all -g
+npx skills add modelstudioai/cli --all -g
 ```
 
 > 需要预先安装 Node.js >= 22.12。
@@ -79,7 +83,10 @@ npx skills add modelstudioai/skills --all -g
 ## 快速开始
 
 ```bash
-# 认证
+# 认证（推荐浏览器登录）
+bl auth login --console
+
+# 或使用 API key 认证
 bl auth login --api-key sk-xxxxx
 
 # 和通义千问对话
@@ -103,9 +110,30 @@ bl advisor recommend --message "qwen-max 和 deepseek-v3 哪个更适合做代
 # 浏览器登录（控制台能力相关命令需要）
 bl auth login --console
 
-# 浏览应用 / 免费额度
+# 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 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       # 非阻塞状态探测（退出码 0/1/3 = 成功/失败/进行中）
+bl finetune capability --model qwen3-8b               # 查询模型支持哪些训练方式
+bl deploy create --model qwen3-8b --name my-svc --plan mu  # 把训练好的模型部署为推理服务
+
+# 浏览应用 / 免费额度 / 用量统计 / 业务空间
 bl app list
-bl usage free --model qwen3-max
+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
 ```
 
 > 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&)
@@ -114,7 +142,7 @@ bl usage free --model qwen3-max
 
 ### DashScope API Key
 
-大部分命令均需要 API Key。前往 [DashScope 控制台](https://bailian.console.aliyun.com/cli?source_channel=key_github&) 获取。
+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
@@ -129,15 +157,15 @@ bl text chat --api-key sk-xxxxx --message "你好"
 
 ### 控制台登录（OAuth）
 
-控制台能力命令（`app list`、`usage free`）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（`app list`、`usage free`、`usage 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。
 
 ```bash
 bl auth login --console
 ```
 
-### 阿里云 AK/SK（仅知识库检索）
+### 阿里云 AK/SK（知识库检索与 Token Plan）
 
-`knowledge retrieve` 命令需要阿里云 AccessKey。前往 [RAM 控制台](https://ram.console.aliyun.com/manage/ak) 获取。
+`knowledge retrieve` 与 `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。
 
 > 建议：创建 RAM 子账号并授予最小权限，避免使用主账号 AK/SK。
 
@@ -154,7 +182,7 @@ export BAILIAN_WORKSPACE_ID=ws-...
 bl config show
 
 # 设置默认值
-bl config set --key region --value us
+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
 
@@ -166,19 +194,19 @@ bl update
 
 ## 相关链接
 
-| 资源                    | 地址                                                              |
-| :---------------------- | :---------------------------------------------------------------- |
-| 阿里云百炼 CLI 官方主页 | https://bailian.console.aliyun.com/cli?source_channel=cli_github& |
-| DashScope API 文档      | https://help.aliyun.com/zh/model-studio/                          |
-| 通义千问模型列表        | https://help.aliyun.com/zh/model-studio/getting-started/models    |
-| 阿里云百炼控制台        | https://bailian.console.aliyun.com/                               |
-| 获取 API Key            | https://bailian.console.aliyun.com/cli?source_channel=key_github& |
-| 获取 AccessKey          | https://ram.console.aliyun.com/manage/ak                          |
+| 资源                    | 地址                                                                                      |
+| :---------------------- | :---------------------------------------------------------------------------------------- |
+| 阿里云百炼 CLI 官方主页 | https://bailian.console.aliyun.com/cli?source_channel=cli_github&                         |
+| DashScope API 文档      | https://help.aliyun.com/zh/model-studio/                                                  |
+| 通义千问模型列表        | https://help.aliyun.com/zh/model-studio/getting-started/models                            |
+| 阿里云百炼控制台        | https://bailian.console.aliyun.com/?source_channel=cli_github                             |
+| 获取 API Key            | https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key |
+| 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_CN.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG_CN.md)。
+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_CN.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING_CN.md)。
+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)。

docs/agents/auth-change.md

@@ -78,9 +78,7 @@ flag 优先 ─→ config 文件 ─→ env var
 
 ### D. main 启动逻辑
 
-- [ ] `packages/cli/src/main.ts:NO_AUTH_SETUP` 列表:
-  - 如果新增的命令"自己管鉴权或不需要鉴权",加进去绕开 ensureApiKey 拦截
-  - 当前清单以 `main.ts:NO_AUTH_SETUP` 为准
+- [ ] 若新增命令**自行处理鉴权**或**不应在入口触发默认 API key 引导**,在对应 `defineCommand` 上设 `skipDefaultApiKeySetup: true`(见 `packages/core/src/types/command.ts`;`packages/cli/src/main.ts` 在 `registry.resolve` 后读取 `command.skipDefaultApiKeySetup`)
 
 ### E. 错误文案
 
@@ -89,7 +87,7 @@ flag 优先 ─→ config 文件 ─→ env var
 
 ### F. 用户面文档
 
-- [ ] `README.md` / `README_CN.md` "Authentication" 段落
+- [ ] `README.md` / `README.zh.md` "Authentication" 段落
 
 ### G. 测试
 

docs/agents/branch-merge-review.md

@@ -50,14 +50,14 @@ git diff --name-only <base>...<head>
 - [ ] **`package.json` 没破坏发布元数据**:`bin` / `exports` / `files` / `inlinedDependencies` 字段任何删除或改名都要单独评估
 - [ ] **公共依赖没被悄悄升级**:catalog / 根 lockfile 改动要列出来
 - [ ] **`package.json` version 没倒退**:目标分支已经更高时(如 main 1.0.3 vs head 1.0.0-beta.1),手动对齐版本号,不要被 head 覆盖
-- [ ] **全局表没冲突**:`registry.ts`、`NO_AUTH_SETUP`(`packages/cli/src/main.ts`)、`ExitCode` 三个全局表新增项不和现有项冲突
+- [ ] **全局表没冲突**:`registry.ts`、`defineCommand` 的 `skipDefaultApiKeySetup`(见 `packages/core/src/types/command.ts`)、`ExitCode` 三处新增项不和现有项冲突
 
 ## 清单 B:用户透出(用户可见的新东西必看)
 
 - [ ] **新命令 / 新 flag** 已同步到用户面文档:
-  - [README.md](README.md) + [README_CN.md](README_CN.md)(中英文都要,常漏 `_CN`)
+  - [README.md](README.md) + [README.zh.md](README.zh.md)(中英文都要,常漏 `_CN`)
   - (SKILL.md 已迁出本仓库,由 `npx add skills` 机制独立维护,不在本仓库 review 范围)
-- [ ] **`bl <cmd> --help`** 文案完整:`description` / `examples` / `apiDocs` 都填了
+- [ ] **`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
 - [ ] **错误信息 / 提示文案**:面向用户的字符串通顺、双语(项目主体是中文场景)
@@ -67,7 +67,7 @@ git diff --name-only <base>...<head>
 - [ ] **改了文件但没补测试**:`git diff --stat <base>...<head> -- '*test*' '*spec*'` 与改动文件清单对照
 - [ ] **新功能埋点同步**:遥测事件名 + 参数 allowlist(参考 main 上的 `feat(telemetry): track console gateway api name in params allowlist` commit)
 - [ ] **环境变量**:新增 / 重命名的 env var 进 README,旧的有没有兼容
-- [ ] **i18n**:`README.md` 改了,`README_CN.md` 同步了吗
+- [ ] **i18n**:`README.md` 改了,`README.zh.md` 同步了吗
 
 ## 输出报告(照模板填)
 
@@ -80,7 +80,7 @@ git diff --name-only <base>...<head>
 解冲突要点(merge 时不要漏):
   - <冲突文件> + <字段/段落> + <怎么取舍>
   ↑ 放"合并那一刻才会出现"的细节,例如 package.json 的 files/scripts/devDependencies 各取并集、
-    NO_AUTH_SETUP 这种全局表两边都加项时不要丢一侧、pnpm-lock.yaml 直接 rm 后 pnpm install 重生等。
+    `skipDefaultApiKeySetup` 这类命令元数据两边都加项时不要丢一侧、pnpm-lock.yaml 直接 rm 后 pnpm install 重生等。
 建议修(可后置):
   - ...
 仅信息(无需动作,告知即可):
@@ -94,11 +94,11 @@ git diff --name-only <base>...<head>
 
 ## 常见漏点(基于历史踩坑)
 
-| 漏点                                                                           | 后果                                                                          |
-| ------------------------------------------------------------------------------ | ----------------------------------------------------------------------------- |
-| `pnpm-workspace.yaml` 把 `packages/*` 收窄成显式列表                           | 合并后目标分支的新子包不再被 workspace 识别,`pnpm install` 看似正常但子包失联 |
-| 源分支 version 比目标分支低,直接 merge 覆盖                                    | npm 上版本号回退,latest tag 错乱                                              |
-| `registry.ts` 注册新命令但忘了 [README](README.md) / [README_CN](README_CN.md) | 用户完全感知不到新功能                                                        |
-| 共享 util 重构(抽公共函数)只改了一处调用方                                     | 其它调用方静默走旧分支,行为分裂                                               |
-| `NO_AUTH_SETUP` 加了不该免登录的命令                                           | 安全风险,用户没登录也能调付费 API                                             |
-| `NO_AUTH_SETUP` / `registry.ts` 这类全局表两边都加项,解冲突时被合掉一侧        | 某个命令突然要求登录 / 某个新命令注册丢失,编译能过、回归不易察觉              |
+| 漏点                                                                            | 后果                                                                          |
+| ------------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
+| `pnpm-workspace.yaml` 把 `packages/*` 收窄成显式列表                            | 合并后目标分支的新子包不再被 workspace 识别,`pnpm install` 看似正常但子包失联 |
+| 源分支 version 比目标分支低,直接 merge 覆盖                                     | npm 上版本号回退,latest tag 错乱                                              |
+| `registry.ts` 注册新命令但忘了 [README](README.md) / [README.zh](README.zh.md)  | 用户完全感知不到新功能                                                        |
+| 共享 util 重构(抽公共函数)只改了一处调用方                                      | 其它调用方静默走旧分支,行为分裂                                               |
+| 不该跳过默认 API key 引导的命令误设 `skipDefaultApiKeySetup: true`              | 安全风险,用户没配置 key 也能调付费 API                                        |
+| `catalog.ts` / `skipDefaultApiKeySetup` 这类元数据两边都加项,解冲突时被合掉一侧 | 某个命令突然要求登录 / 某个新命令注册丢失,编译能过、回归不易察觉              |

docs/agents/changelog-write.md

@@ -10,7 +10,7 @@
 更新仓库内的两份文件,英文优先,中文同步:
 
 - [`CHANGELOG.md`](../../CHANGELOG.md)
-- [`CHANGELOG_CN.md`](../../CHANGELOG_CN.md)
+- [`CHANGELOG.zh.md`](../../CHANGELOG.zh.md)
 
 新版本条目插在文件顶部"## [X.Y.Z] - YYYY-MM-DD"位置,旧版本依次向下保留。两份文件保持一一对应——任何条目只在一份里出现,另一份漏写,视为错误。
 
@@ -129,7 +129,7 @@ git show <commit> --stat
 
 ### 8. 写完后给用户过一遍再写入文件
 
-**不要直接编辑 `CHANGELOG.md` / `CHANGELOG_CN.md`**。先把中英两份草稿都贴回对话里,让用户:
+**不要直接编辑 `CHANGELOG.md` / `CHANGELOG.zh.md`**。先把中英两份草稿都贴回对话里,让用户:
 
 - 增删条目
 - 调整措辞(中英、术语)
@@ -183,4 +183,4 @@ git show <commit> --stat
 | [publish.md](publish.md) | 发布流程:自检 / 构建 / npm publish（CI 驱动） |
 | 本文档                   | 发版后写说明:面向用户的 release notes         |
 
-两者顺序:`publish.md` → npm publish → 本文档(更新 `CHANGELOG.md` + `CHANGELOG_CN.md`)→ 推到 GitHub。
+两者顺序:`publish.md` → npm publish → 本文档(更新 `CHANGELOG.md` + `CHANGELOG.zh.md`)→ 推到 GitHub。

docs/agents/cli-e2e-tests.md

@@ -60,7 +60,7 @@ describe.skipIf(<ready>)("e2e: <topic>（DashScope …）", () => {
 ## 新增 command 检查清单
 
 - [ ] `commands/catalog.ts` 登记 + `tests/e2e/<topic>.e2e.test.ts`（新建或扩展）
-- [ ] 若改了 `usage` / `options` / `examples`,跑 `pnpm --filter bailian-cli run generate:reference` 更新 `tools/generated/reference/`(本仓库 gitignore)
+- [ ] 若改了 `usage` / `options` / `examples`,跑 `pnpm --filter bailian-cli run generate:reference` 更新 `skills/bailian-cli/reference/` 并提交
 - [ ] 顶层：分组 help + 子命令 `--help`（多子命令则各一条 help）
 - [ ] 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

docs/agents/command-add-remove.md

@@ -27,20 +27,20 @@
 命令元数据以 **`catalog.ts` 为单一登记处**;`registry.ts` 只负责解析与打印 help,不再内嵌命令表或手写 Resources 列表。
 
 ```
-commands/<...>.ts   defineCommand({ name, description, usage, options, examples, apiDocs?, run })
+commands/<...>.ts   defineCommand({ name, description, usage, options, examples, run })
         ↓
 commands/catalog.ts   export const commands: Record<string, Command>
         ↓
    ┌────┴────┬──────────────────────┬─────────────────────┐
    ↓         ↓                      ↓                     ↓
 registry.ts  main.ts      tools/generate-reference.ts   export-schema.ts
-(解析/help)  (入口)       → tools/generated/reference/index.md + <group>.md
+(解析/help)  (入口)       → skills/bailian-cli/reference/index.md + <group>.md
 ```
 
 - **`packages/cli/src/commands/catalog.ts`**: `import` 命令模块 + `"<path>": handler` 映射;**不** `import registry.ts`(避免构建时循环依赖)
 - **`packages/cli/src/commands/index.ts`**: `export { commands } from "./catalog.ts"`(给包内 re-export 用)
 - **`packages/cli/src/registry.ts`**: `import { commands } from "./commands/catalog.ts"`,建树、`resolve`、`printHelp`;Commands / Global Flags 从 `Command` 元数据与 `GLOBAL_OPTIONS` **动态生成**
-- **`tools/generate-reference.ts`**: build 前读 `catalog.ts`,写 `tools/generated/reference/index.md`(索引) + `tools/generated/reference/<一级命令>.md`(详情,勿手改)。该目录被 gitignore,产物供未来的 `npx add skills` 安装机制消费
+- **`tools/generate-reference.ts`**: pre-commit / `pnpm run sync:skill-assets` 时读 `catalog.ts`,写 `skills/bailian-cli/reference/index.md`(索引) + `skills/bailian-cli/reference/<一级命令>.md`(详情,勿手改)。该目录**纳入 git**,随 `npx skills add modelstudioai/cli` 分发
 
 已删除、勿再引用:`commands/help.ts`、`registry.ts` 内联 `new CommandRegistry({...})`、`printRootHelp` 手写命令行。
 
@@ -53,15 +53,14 @@ registry.ts  main.ts      tools/generate-reference.ts   export-schema.ts
   - 增删 `import xxx from "./.../xxx.ts"`
   - 在 `export const commands` 里增删 `"<group> <action>": xxx`(key 与 `defineCommand({ name })` 一致)
 - [ ] **不要**在 `registry.ts` 里重复登记命令(已从 catalog 读取)
-- [ ] 命令需在 `bl help` / `reference/` 展示 API 文档链接时,在 `defineCommand` 里设 `apiDocs`(相对路径);help 与 reference 均从此字段生成
-- [ ] 如果命令需要鉴权之外的特殊路径,看 `packages/cli/src/main.ts` 的 `NO_AUTH_SETUP`
+- [ ] 如果命令需要跳过入口的默认 DashScope API key 引导(`ensureApiKey`),在对应 `defineCommand` 上设 `skipDefaultApiKeySetup: true`(字段定义见 `packages/core/src/types/command.ts`;`main.ts` 根据已解析的 `command` 读取)
 - [ ] **`config/export-schema.ts`**: 若新命令不适合作为 agent tool,评估是否加入 `SKIP_PREFIXES`;该文件在 `run()` 内 `import("../catalog.ts")`,勿顶层 import catalog 以免循环依赖
 
 ### B. 文档层
 
-- [ ] 运行 `pnpm --filter bailian-cli run generate:reference`(或 `build`),刷新 `tools/generated/reference/` 下生成文件(本仓库 gitignore,仅供本地校验和未来 skill 安装机制消费)
-- [ ] `README.md` / `README_CN.md`: Quick Start、命令一览(用户向,与 help 对齐即可)
-- [ ] SKILL.md 已搬出本仓库(由 `npx add skills` 机制分发),本仓库不再维护
+- [ ] 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
+- [ ] `README.md` / `README.zh.md`: Quick Start、命令一览(用户向,与 help 对齐即可)
+- [ ] `skills/bailian-cli/SKILL.md`: 若安装说明或能力边界有变,同步更新
 
 ### C. 测试层
 
@@ -73,14 +72,14 @@ registry.ts  main.ts      tools/generate-reference.ts   export-schema.ts
 - [ ] 全仓 grep **旧命令名字符串**,确保以下位置全部更新:
   - `catalog.ts` 的 key
   - error hints(cli 层)
-  - `tools/generated/reference/`(重建后检查;本仓库 gitignore)
+  - `skills/bailian-cli/reference/`(重建后检查并提交)
   - README 示例
   - 测试断言
 
 ## 完成后自查
 
 ```sh
-pnpm --filter bailian-cli run generate:reference   # reference/ 与 catalog 一致
+pnpm run sync:skill-assets   # reference/ + SKILL metadata.version 与 catalog / package.json 一致
 node packages/cli/src/main.ts <new-command> --help
 node packages/cli/src/main.ts                        # 根 help 列表含新命令
 vp test packages/cli/tests/e2e/<topic>.e2e.test.ts   # 相关 e2e
@@ -89,6 +88,6 @@ vp test packages/cli/tests/e2e/<topic>.e2e.test.ts   # 相关 e2e
 ## 常见漏点
 
 - ✗ 只改了命令文件,忘了 **`catalog.ts`** → 命令不存在或 help 里没有
-- ✗ 手改 **`tools/generated/reference/*.md`** → 下次 build 被覆盖;应改 `defineCommand` 后重新 generate
+- ✗ 手改 **`skills/bailian-cli/reference/*.md`** → 下次 generate 被覆盖;应改 `defineCommand` 后重新 generate 并提交
 - ✗ 在 `export-schema.ts` 顶层 `import catalog` → 可能与 registry 循环依赖
 - ✗ 单 action 的子组是反模式,新增时优先拍平为两级

docs/agents/command-flag-change.md

@@ -28,8 +28,8 @@
 
 ### C. 文档层
 
-- [ ] `README.md` / `README_CN.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 `tools/generated/reference/` 与命令一致(本仓库 gitignore,勿手改;SKILL.md 已迁出本仓库)
+- [ ] `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)
 
 ### D. 测试层
 

docs/agents/config-add.md

@@ -54,7 +54,7 @@ config 文件     ─┘
 
 ### E. 文档
 
-- [ ] `README.md` / `README_CN.md` 的 env var 表格
+- [ ] `README.md` / `README.zh.md` 的 env var 表格
 
 ### F. 测试
 

docs/agents/lint-toolchain.md

@@ -40,6 +40,7 @@
 
 - [ ] `.vite-hooks/pre-commit` 改动后,`pnpm install` 重新软链(走 `prepare: vp config`)
 - [ ] 增加 hook 时,确认在干净 clone 后能自动激活
+- [ ] pre-commit 会跑 `pnpm run sync:skill-assets`(先 build core,再 `generate:reference` + `sync:skill-version`)并 `git add` skill 资产,最后 `vp staged`
 
 ### F. CI / 发版工具
 

docs/agents/model-add-remove.md

@@ -25,11 +25,11 @@
 
 ### 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 `tools/generated/reference/<group>.md`(本仓库 gitignore;SKILL.md 由独立的 `npx add skills` 仓库维护,本仓库不再含)
+- [ ] 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
 
 ### D. 用户面文档
 
-- [ ] `README.md` / `README_CN.md`:
+- [ ] `README.md` / `README.zh.md`:
   - Quick Start 示例如使用了具体型号,确认仍可用
   - 顶部 introduction 段落如提到"Qwen-Omni"等品牌名,无需变(模型代号变化不算品牌变)
 

docs/agents/publish.md

@@ -67,9 +67,15 @@ node tools/release/publish-channel.mjs --channel test --dry-run
 - [ ] `packages/cli/package.json` 和 `packages/core/package.json` 已升到目标版本
 - [ ] pre-release 格式正确（`1.0.0-beta.0` / `1.0.0-rc.1`，**不要直接用 `1.0.0` 当 beta**）
 
+### CHANGELOG（仅 stable）
+
+- [ ] `CHANGELOG.md` 和 `CHANGELOG.zh.md` 都已新增目标版本条目，中英文一一对应
+- [ ] 分类标题用 Keep a Changelog 规范的 `Added` / `Changed` / `Deprecated` / `Removed` / `Fixed` / `Security`（中文版对应 `新增` / `变更` / `已弃用` / `已移除` / `修复` / `安全`），**不要自创 `Improved` / `优化` 等规范外分类**
+- [ ] 条目日期与发版日期一致
+
 ### 用户面文档
 
-- [ ] `README.md` / `README_CN.md` 的 Quick Start 命令仍能跑通
+- [ ] `README.md` / `README.zh.md` 的 Quick Start 命令仍能跑通
 - [ ] README 的 Node.js 徽章版本与 `cli/package.json.engines.node` 一致
 - [ ] README 宣传的 bin 名称在 `cli/package.json.bin` 都真的注册
 - [ ] `LICENSE` 文件存在（根 + cli + core 各一份）
@@ -81,11 +87,12 @@ node tools/release/publish-channel.mjs --channel test --dry-run
 
 ## 常见漏点（基于历史踩坑）
 
-| 漏点                                                  | 后果                                               |
-| ----------------------------------------------------- | -------------------------------------------------- |
-| cli 升版号但 core 没升                                | check.mjs 会拦下                                   |
-| `1.0.0` 当 beta 直接发                                | 占了 `latest` tag，所有用户被强升，撤回成本极高    |
-| README 写的 bin 名实际 `package.json.bin` 没注册      | 用户复制命令报 `command not found`                 |
-| Node 徽章 `>=18`、engines `>=22.12` 不一致            | 用户在 Node 18 上 `npm i` 被 engine 警告或直接失败 |
-| 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      |
+| 漏点                                                     | 后果                                               |
+| -------------------------------------------------------- | -------------------------------------------------- |
+| cli 升版号但 core 没升                                   | check.mjs 会拦下                                   |
+| 发版漏更 CHANGELOG，或分类写成规范外的 `优化`/`Improved` | 用户看不到本次变更，分类与历史不一致               |
+| `1.0.0` 当 beta 直接发                                   | 占了 `latest` tag，所有用户被强升，撤回成本极高    |
+| README 写的 bin 名实际 `package.json.bin` 没注册         | 用户复制命令报 `command not found`                 |
+| Node 徽章 `>=18`、engines `>=22.12` 不一致               | 用户在 Node 18 上 `npm i` 被 engine 警告或直接失败 |
+| 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      |

docs/agents/url-change.md

@@ -46,8 +46,8 @@ 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)
 
-- [ ] `tools/generated/reference/` 各 `<group>.md` 中 API/控制台 URL(`generate:reference` 重建后核对;本仓库 gitignore)
-- [ ] `README.md` / `README_CN.md` 中所有 URL
+- [ ] `skills/bailian-cli/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
 

docs/plans/finetune-deploy-mvp.md

@@ -0,0 +1,438 @@
+# 模型训练 + 数据集 + 部署：最小闭环 CLI 设计
+
+> 目标：一个 Qwen 文本模型 SFT 训练、数据集上传、模型部署的端到端最小链路。
+
+---
+
+## 一、命令概览
+
+| 优先级 | 命令                                | 映射 API                                      | 用途                            |
+| ------ | ----------------------------------- | --------------------------------------------- | ------------------------------- |
+| P0     | `bl dataset upload <path>`          | `POST /api/v1/files`                          | 上传训练数据(含本地格式校验)    |
+| P0     | `bl finetune create`                | `POST /api/v1/fine-tunes`                     | 创建 SFT 训练任务(预填默认超参) |
+| P0     | `bl finetune status <job_id>`       | `GET /api/v1/fine-tunes/{job_id}`             | 查询训练状态                    |
+| P0     | `bl deploy create`                  | `POST /api/v1/deployments`                    | 部署训练好的模型                |
+| P1     | `bl finetune logs <job_id>`         | `GET /api/v1/fine-tunes/{job_id}/logs`        | 拉取训练日志                    |
+| P1     | `bl finetune checkpoints <job_id>`  | `GET /api/v1/fine-tunes/{job_id}/checkpoints` | 查看/挑选 Checkpoint            |
+| P1     | `bl deploy status <deployed_model>` | `GET /api/v1/deployments/{deployed_model}`    | 查询部署状态                    |
+| P1     | `bl deploy delete <deployed_model>` | `DELETE /api/v1/deployments/{deployed_model}` | 下线部署                        |
+| P1     | `bl infer --model <deployed_model>` | 复用 `text chat` 通路                         | 调用已部署模型                  |
+
+---
+
+## 二、P0 命令详细设计
+
+### 2.1 `bl dataset upload`
+
+**定位：** 上传训练数据文件到百炼平台，获取 `file_id` 供训练任务引用。
+
+#### CLI 签名
+
+```
+bl dataset upload <path> [--purpose fine-tune] [--validate] [--no-validate]
+```
+
+| Flag            | 必填 | 默认值      | 说明                           |
+| --------------- | ---- | ----------- | ------------------------------ |
+| `<path>`        | 是   | —           | 本地文件路径（.jsonl 或 .zip） |
+| `--purpose`     | 否   | `fine-tune` | 文件用途标签                   |
+| `--validate`    | 否   | `true`      | 上传前执行本地格式校验         |
+| `--no-validate` | 否   | —           | 跳过本地校验                   |
+
+#### 本地格式校验规则（提交前拦截）
+
+校验逻辑在 `packages/core` 实现（纯函数），CLI 调用后展示错误：
+
+1. **文件格式检查**：仅允许 `.jsonl` 和 `.zip`（zip 内根目录必须有 `data.jsonl`）
+2. **JSONL 逐行校验**：
+   - 每行可被 `JSON.parse`
+   - 顶层必须包含 `messages` 数组
+   - `messages` 中每项必须包含 `role`（枚举：`system` | `user` | `assistant`）和 `content`（非空字符串）
+   - 至少包含一条 `user` + 一条 `assistant` 消息
+3. **数量校验**：SFT 训练至少需要上千条数据（给出 warning 而非 hard fail，阈值建议 ≥ 10 条 hard fail）
+4. **文件体积**：≤ 300MB
+
+#### 校验失败输出示例
+
+```
+✗ Validation failed:
+
+  Line 3: missing "messages" field
+  Line 7: role "bot" is not valid (expected: system | user | assistant)
+  Line 12: "content" is empty string
+
+Fix 3 errors above and retry.
+```
+
+#### API 调用
+
+```
+POST https://dashscope.aliyuncs.com/api/v1/files
+Content-Type: multipart/form-data
+Authorization: Bearer <api-key>
+
+Body:
+  files: <binary>
+  purpose: "fine-tune"
+
+Response 200:
+{
+  "id": "file-xxxx",
+  "bytes": 12345,
+  "filename": "train.jsonl",
+  "purpose": "fine-tune",
+  "created_at": 1700000000
+}
+```
+
+#### 输出
+
+- 默认 text：`✓ Uploaded file-xxxx (12.3 KB) — use this ID in bl finetune create`
+- `--output json`：完整 response body
+- `--quiet`：仅输出 `file-xxxx`
+
+---
+
+### 2.2 `bl finetune create`
+
+**定位：** 创建一个 SFT 训练任务。核心设计原则——**预填合理默认超参 + 提交前二次确认**，降低 OOM/超参不合理导致的训练失败率。
+
+#### CLI 签名
+
+```
+bl finetune create --model <model> --data <file_id> [hyperparams...]
+```
+
+| Flag                | 必填 | 默认值       | 说明                                         |
+| ------------------- | ---- | ------------ | -------------------------------------------- |
+| `--model`           | 是   | —            | 基座模型（如 `qwen3-8b`, `qwen3-14b`）       |
+| `--data`            | 是   | —            | 训练数据 file_id（bl dataset upload 返回值） |
+| `--validation-data` | 否   | —            | 验证数据 file_id                             |
+| `--epochs`          | 否   | 3            | 训练轮次 (n_epochs)                          |
+| `--batch-size`      | 否   | 按模型自动选 | 批大小                                       |
+| `--lr`              | 否   | 按模型自动选 | 学习率 (learning_rate_multiplier)            |
+| `--warmup-ratio`    | 否   | 0.1          | warmup 比例                                  |
+| `--suffix`          | 否   | —            | 输出模型后缀名                               |
+| `--yes` / `-y`      | 否   | —            | 跳过确认直接提交                             |
+
+#### 预填默认超参策略
+
+| 基座模型   | batch_size | lr_multiplier | n_epochs | 备注             |
+| ---------- | ---------- | ------------- | -------- | ---------------- |
+| qwen3-8b   | 4          | 1e-5          | 3        | 小模型可大 batch |
+| qwen3-14b  | 2          | 5e-6          | 3        | 中模型防 OOM     |
+| qwen3-32b+ | 1          | 2e-6          | 2        | 大模型保守设置   |
+
+> 以上为建议默认值，用户显式传参时覆盖。具体映射表在 `packages/core/src/finetune/defaults.ts` 维护。
+
+#### 提交前交互确认
+
+非 `--yes` 模式下，显示任务摘要等待确认：
+
+```
+┌─ Fine-tune Job Summary ──────────────────────┐
+│ Model:        qwen3-8b                        │
+│ Training:     file-abc123 (2,048 samples)     │
+│ Validation:   (none)                          │
+│ Epochs:       3                               │
+│ Batch size:   4                               │
+│ LR:           1e-5                            │
+│ Warmup:       0.1                             │
+│ Suffix:       my-assistant                    │
+│                                               │
+│ Estimated cost: ~¥XX (based on token count)   │
+└───────────────────────────────────────────────┘
+Proceed? [Y/n]
+```
+
+#### API 调用
+
+```
+POST https://dashscope.aliyuncs.com/api/v1/fine-tunes
+Authorization: Bearer <api-key>
+Content-Type: application/json
+
+{
+  "model": "qwen3-8b",
+  "training_file_ids": ["file-abc123"],
+  "validation_file_ids": [],
+  "hyper_parameters": {
+    "n_epochs": 3,
+    "batch_size": 4,
+    "learning_rate": "1e-5",
+    "warmup_ratio": 0.1
+  },
+  "suffix": "my-assistant"
+}
+
+Response 200:
+{
+  "job_id": "ft-xxxx",
+  "status": "PENDING",
+  "model": "qwen3-8b",
+  "created_at": "2025-01-01T00:00:00Z",
+  "training_file_ids": ["file-abc123"],
+  "hyper_parameters": {...},
+  "trained_model": null
+}
+```
+
+#### 输出
+
+- text：`✓ Fine-tune job ft-xxxx created (PENDING). Track with: bl finetune status ft-xxxx`
+- json：完整 response body
+- quiet：`ft-xxxx`
+
+---
+
+### 2.3 `bl finetune status`
+
+**定位：** 查询训练任务状态，支持 `--wait` 轮询模式。
+
+#### CLI 签名
+
+```
+bl finetune status <job_id> [--wait] [--interval <seconds>]
+```
+
+| Flag         | 必填 | 默认值 | 说明             |
+| ------------ | ---- | ------ | ---------------- |
+| `<job_id>`   | 是   | —      | 任务 ID          |
+| `--wait`     | 否   | —      | 持续轮询直到终态 |
+| `--interval` | 否   | 30     | 轮询间隔(秒)     |
+
+#### 状态机
+
+```
+PENDING → RUNNING → SUCCEEDED
+                  ↘ FAILED
+```
+
+#### 输出（text 模式）
+
+单次查询：
+
+```
+Job:     ft-xxxx
+Status:  RUNNING (elapsed 12m)
+Model:   qwen3-8b
+Output:  (pending)
+```
+
+`--wait` 模式（spinner + 实时刷新）：
+
+```
+⠋ ft-xxxx RUNNING [14:32 elapsed]
+✓ ft-xxxx SUCCEEDED — trained model: qwen3-8b:ft-xxxx-20250101
+  Deploy with: bl deploy create --model qwen3-8b:ft-xxxx-20250101
+```
+
+失败时：
+
+```
+✗ ft-xxxx FAILED
+  Error: OutOfMemory — try reducing --batch-size or using a smaller model
+```
+
+---
+
+### 2.4 `bl deploy create`
+
+**定位：** 将训练好的模型（或 checkpoint）部署为可调用的推理服务。
+
+#### CLI 签名
+
+```
+bl deploy create --model <model_name> [--plan <plan>] [--capacity <n>]
+```
+
+| Flag         | 必填 | 默认值     | 说明                                            |
+| ------------ | ---- | ---------- | ----------------------------------------------- |
+| `--model`    | 是   | —          | 待部署模型名称（finetune 产出的 trained_model） |
+| `--plan`     | 否   | `standard` | 部署方案                                        |
+| `--capacity` | 否   | 依 plan    | 并发容量                                        |
+| `--wait`     | 否   | —          | 等待部署就绪                                    |
+
+#### API 调用
+
+```
+POST https://dashscope.aliyuncs.com/api/v1/deployments
+Authorization: Bearer <api-key>
+Content-Type: application/json
+
+{
+  "model_name": "qwen3-8b:ft-xxxx-20250101",
+  "plan": "standard",
+  "capacity": 2
+}
+
+Response 200:
+{
+  "deployed_model": "qwen3-8b-ft-xxxx",
+  "model_name": "qwen3-8b:ft-xxxx-20250101",
+  "status": "PENDING",
+  "created_at": "..."
+}
+```
+
+#### 输出
+
+```
+✓ Deployment created: qwen3-8b-ft-xxxx (PENDING)
+  Once RUNNING, call with: bl text chat --model qwen3-8b-ft-xxxx
+  Check status: bl deploy status qwen3-8b-ft-xxxx
+```
+
+---
+
+## 三、P1 命令简要设计
+
+### 3.1 `bl finetune logs <job_id>`
+
+流式输出训练日志，支持 `--follow`（类似 `tail -f`）。输出 loss/step/epoch 信息。
+
+### 3.2 `bl finetune checkpoints <job_id>`
+
+列出可选 checkpoint（step, loss, eval metrics），支持 `--output json` 供脚本使用。可配合 `bl deploy create --model <checkpoint_model>` 部署指定 checkpoint。
+
+### 3.3 `bl deploy status <deployed_model>`
+
+查询部署状态及资源信息（PENDING → RUNNING → STOPPED/FAILED）。
+
+### 3.4 `bl deploy delete <deployed_model>`
+
+下线部署。需部署处于 RUNNING/STOPPED/FAILED 状态。交互确认或 `--yes` 跳过。
+
+### 3.5 `bl infer --model <deployed_model>`
+
+实际可复用已有 `bl text chat --model <deployed_model>` 通路，作为别名/快捷方式。P1 考虑是否有独立存在必要。
+
+---
+
+## 四、代码架构方案
+
+按照 monorepo 分层约定（core 纯逻辑 / cli 是 UI）：
+
+### packages/core 新增模块
+
+```
+packages/core/src/
+├── finetune/
+│   ├── index.ts          # re-export
+│   ├── api.ts            # createFineTune, getFineTune, getFineTuneLogs, getCheckpoints
+│   ├── defaults.ts       # 模型 → 默认超参映射表
+│   └── types.ts          # FineTuneJob, HyperParameters, CheckpointInfo 类型
+├── dataset/
+│   ├── index.ts
+│   ├── upload.ts         # uploadDataset (multipart)
+│   ├── validate.ts       # validateJsonl (纯函数,逐行校验)
+│   └── types.ts          # DatasetFile, ValidationError 类型
+└── deploy/
+    ├── index.ts
+    ├── api.ts            # createDeployment, getDeployment, deleteDeployment
+    └── types.ts          # Deployment, DeploymentStatus 类型
+```
+
+### packages/cli 新增命令
+
+```
+packages/cli/src/commands/
+├── dataset/
+│   └── upload.ts         # bl dataset upload
+├── finetune/
+│   ├── create.ts         # bl finetune create
+│   ├── status.ts         # bl finetune status
+│   ├── logs.ts           # bl finetune logs
+│   └── checkpoints.ts    # bl finetune checkpoints
+└── deploy/
+    ├── create.ts         # bl deploy create
+    ├── status.ts         # bl deploy status
+    └── delete.ts         # bl deploy delete
+```
+
+---
+
+## 五、关键设计决策
+
+### 5.1 数据格式校验放在 CLI 侧（提交前拦截）
+
+训练失败 TOP 原因中"数据格式错误"占比高。与其等服务端 10 分钟后返回 FAILED，不如 CLI 本地秒级校验：
+
+- **validate.ts** 是纯函数，接收 ReadableStream/Buffer，返回 `ValidationError[]`
+- CLI 在 `dataset upload` 默认执行校验，`--no-validate` 允许跳过
+- 未来可扩展为独立命令 `bl dataset validate <path>`
+
+### 5.2 超参预填 + 确认而非强制
+
+- core 维护 `defaults.ts` 映射：`model → { batch_size, lr, epochs }`
+- CLI `finetune create` 未指定超参时自动填入
+- 提交前展示完整参数面板（非 --yes 模式），避免"我以为用了默认但其实没传"
+
+### 5.3 费用感知（P1+）
+
+- 图像/语音/视频训练费用远高于文本。MVP 阶段（Qwen 文本 SFT）费用可控
+- 后续扩展多模态时，在 confirm panel 中强化费用估算提示
+- `bl quota check` 已存在，可在 `finetune create` 内部集成余额预检
+
+### 5.4 `bl infer` 是否独立存在
+
+建议 P1 阶段**不新增** `bl infer`，而是让 `bl text chat --model <deployed_model>` 直接工作。部署完成后的引导文案中指明这个用法即可。减少命令膨胀。
+
+---
+
+## 六、最小闭环用户操作流
+
+```bash
+# 1. 准备数据 → 上传（含校验）
+bl dataset upload ./train.jsonl
+# ✓ Uploaded file-abc123 (5.2 MB)
+
+# 2. 创建训练任务（自动预填超参）
+bl finetune create --model qwen3-8b --data file-abc123
+# Shows summary panel → confirm → ✓ Job ft-xxxx created
+
+# 3. 等待训练完成
+bl finetune status ft-xxxx --wait
+# ⠋ RUNNING [23:15] → ✓ SUCCEEDED: qwen3-8b:ft-xxxx-20250601
+
+# 4. 部署模型
+bl deploy create --model qwen3-8b:ft-xxxx-20250601 --wait
+# ✓ Deployed: qwen3-8b-ft-xxxx (RUNNING)
+
+# 5. 调用模型
+bl text chat --model qwen3-8b-ft-xxxx "你好，介绍一下你自己"
+# (正常推理输出)
+```
+
+---
+
+## 七、实现顺序建议
+
+```
+Phase 1 (P0 — 最小闭环):
+  core: dataset/validate.ts → dataset/upload.ts → finetune/api.ts → deploy/api.ts
+  cli:  dataset upload → finetune create → finetune status → deploy create
+  测试: 单元测试 validate.ts + e2e dry-run + 真实 API 端到端一次
+
+Phase 2 (P1 — 可观测性):
+  finetune logs → finetune checkpoints → deploy status → deploy delete
+  费用估算集成
+
+Phase 3 (后续):
+  bl dataset validate (独立命令)
+  bl dataset list (查看已上传)
+  bl finetune list (查看历史任务)
+  多模态 SFT 支持（图像/视频数据格式校验扩展）
+```
+
+---
+
+## 八、风险与 TODO
+
+| 风险点            | 影响              | 缓解措施                                      |
+| ----------------- | ----------------- | --------------------------------------------- |
+| OOM 训练失败      | 用户浪费时间/金钱 | 保守默认超参 + batch_size 自适应模型大小      |
+| 数据格式错误      | 训练启动后才失败  | 本地校验拦截，启动秒级反馈                    |
+| 部署等待时间长    | 用户困惑          | `--wait` + 预估时间提示                       |
+| 费用超预期        | 账号欠费          | confirm panel 预估费用（P1 集成 quota check） |
+| API endpoint 变动 | 调用失败          | 端点集中管理在 core/client/endpoints.ts       |

package.json

@@ -16,8 +16,10 @@
     "ready": "vp check && vp run -r test && vp run -r build",
     "prepare": "vp config",
     "check": "vp check",
+    "sync:skill-assets": "pnpm --filter \"bailian-cli^...\" run build && pnpm --filter bailian-cli run generate:reference && pnpm --filter bailian-cli run sync:skill-version",
     "dev": "pnpm -F bailian-cli-core dev",
     "bl": "pnpm -F bailian-cli dev",
+    "kscli": "pnpm -F knowledge-studio-cli dev",
     "test": "vp test",
     "release:check": "node tools/release/check.mjs",
     "wiki:crawl": "node tools/wiki-crawler/index.mjs",

packages/cli/README.md

@@ -9,7 +9,7 @@
 [![TypeScript](https://img.shields.io/badge/TypeScript-strict-3178c6)](https://www.typescriptlang.org)
 [![License](https://img.shields.io/badge/license-Apache%202.0-blue)](LICENSE)
 
-[Aliyun Model Studio CLI Site](https://bailian.console.aliyun.com/cli?source_channel=cli_github&) · [中文文档](https://github.com/modelstudioai/cli/blob/main/README_CN.md) · [API Documentation](https://help.aliyun.com/zh/model-studio/) · [Get API Key](https://bailian.console.aliyun.com/cli?source_channel=key_github&)
+[Aliyun Model Studio CLI Site](https://bailian.console.aliyun.com/cli?source_channel=cli_github&) · [中文文档](https://github.com/modelstudioai/cli/blob/main/README.zh.md) · [API Documentation](https://help.aliyun.com/zh/model-studio/) · [Get API Key](https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key)
 
 ---
 
@@ -27,15 +27,19 @@ Equip your AI Agent out-of-the-box with these capabilities, composable across co
 - **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-1.0 series: text-/image-/reference-to-video and natural-language video editing (up to 9-image reference)
+- **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
+
+> **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
-- **Console capabilities** — Browse Bailian apps (`app list`) and check free-tier quota (`usage free`)
+- **Fine-tuning & deployment** — Upload datasets, create SFT/LoRA/DPO/CPT jobs (`finetune create`), probe job status non-blockingly (`finetune watch`), query per-model training capability (`finetune capability`), and deploy trained models as endpoints (`deploy create`)
+- **Console capabilities** — Browse Bailian apps (`app list`), 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
@@ -51,7 +55,7 @@ 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 1.0**, Aliyun Model Studio's text-/image-/reference-to-video generation model
+- **[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
 - **[spark-video Skill](https://github.com/JohnKeating1997/spark-video)** — handles scene decomposition, storyboarding, shot continuity, and final stitching
 
 ### The single prompt
@@ -64,7 +68,7 @@ A complete **2-minute, 16:9 cinematic short film** — produced end-to-end from
 
 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 1.0** in parallel.
+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.
 
 No timeline scrubbing. No frame-by-frame editing. Just one sentence → one video.
@@ -73,7 +77,7 @@ No timeline scrubbing. No frame-by-frame editing. Just one sentence → one vide
 
 ```bash
 npm install -g bailian-cli
-npx skills add modelstudioai/skills --all -g
+npx skills add modelstudioai/cli --all -g
 ```
 
 > Requires Node.js >= 22.12.
@@ -108,9 +112,30 @@ bl advisor recommend --message "qwen-max vs deepseek-v3 for code generation"
 # Browser login (required for console capability commands)
 bl auth login --console
 
-# Browse apps / free-tier quota
+# Fine-tune & deploy — a one-shot train-to-serve workflow
+bl dataset upload --file ./train.jsonl                 # Upload a .jsonl dataset (validated first)
+bl finetune 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 status probe (exit 0/1/3 = done/failed/running)
+bl finetune capability --model qwen3-8b               # Which training types a model supports
+bl deploy create --model qwen3-8b --name my-svc --plan mu  # Deploy the trained model as an endpoint
+
+# Browse apps / free-tier quota / usage statistics / workspaces
 bl app list
-bl usage free --model qwen3-max
+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
 ```
 
 > More examples and scenarios: [Aliyun Model Studio CLI Site](https://bailian.console.aliyun.com/cli?source_channel=cli_github&)
@@ -119,7 +144,7 @@ bl usage free --model qwen3-max
 
 ### DashScope API Key
 
-Required for most commands. Get your key from the [DashScope Console](https://bailian.console.aliyun.com/cli?source_channel=key_github&).
+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
@@ -134,15 +159,15 @@ bl text chat --api-key sk-xxxxx --message "Hello"
 
 ### Console Login (OAuth)
 
-Required for console capability commands (`app list`, `usage free`). Opens the Bailian console in your browser to sign in.
+Required for console capability commands (`app list`, `usage free`, `usage stats`, `workspace list`, `quota list/request/check/history`). Opens the Bailian console in your browser to sign in.
 
 ```bash
 bl auth login --console
 ```
 
-### Alibaba Cloud AK/SK (Knowledge Base only)
+### Alibaba Cloud AK/SK (Knowledge Base & Token Plan)
 
-Required for `knowledge retrieve`. Get your AccessKey from [RAM Console](https://ram.console.aliyun.com/manage/ak).
+Required for `knowledge retrieve` and the `token-plan` command group. Get your AccessKey from [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.
 
@@ -159,7 +184,7 @@ export BAILIAN_WORKSPACE_ID=ws-...
 bl config show
 
 # Set defaults
-bl config set --key region --value us
+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
 
@@ -171,14 +196,14 @@ Config file location: `~/.bailian/config.json`
 
 ## Links
 
-| Resource                     | URL                                                               |
-| :--------------------------- | :---------------------------------------------------------------- |
-| Aliyun Model Studio CLI Site | https://bailian.console.aliyun.com/cli?source_channel=cli_github& |
-| DashScope API Docs           | https://help.aliyun.com/zh/model-studio/                          |
-| Qwen Model List              | https://help.aliyun.com/zh/model-studio/getting-started/models    |
-| Aliyun Model Studio Console  | https://bailian.console.aliyun.com/                               |
-| Get API Key                  | https://bailian.console.aliyun.com/cli?source_channel=key_github& |
-| Get AccessKey                | https://ram.console.aliyun.com/manage/ak                          |
+| Resource                     | URL                                                                                       |
+| :--------------------------- | :---------------------------------------------------------------------------------------- |
+| Aliyun Model Studio CLI Site | https://bailian.console.aliyun.com/cli?source_channel=cli_github&                         |
+| DashScope API Docs           | https://help.aliyun.com/zh/model-studio/                                                  |
+| Qwen Model List              | https://help.aliyun.com/zh/model-studio/getting-started/models                            |
+| Aliyun Model Studio Console  | https://bailian.console.aliyun.com/?source_channel=cli_github                             |
+| Get API Key                  | https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key |
+| Get AccessKey                | https://ram.console.aliyun.com/manage/ak                                                  |
 
 ## Changelog
 

packages/cli/README.zh.md

@@ -9,7 +9,7 @@
 [![TypeScript](https://img.shields.io/badge/TypeScript-strict-3178c6)](https://www.typescriptlang.org)
 [![License](https://img.shields.io/badge/license-Apache%202.0-blue)](LICENSE)
 
-[阿里云百炼 CLI 官方主页](https://bailian.console.aliyun.com/cli?source_channel=cli_github&) · [English](https://github.com/modelstudioai/cli/blob/main/README.md) · [API 文档](https://help.aliyun.com/zh/model-studio/) · [获取 API Key](https://bailian.console.aliyun.com/cli?source_channel=key_github&)
+[阿里云百炼 CLI 官方主页](https://bailian.console.aliyun.com/cli?source_channel=cli_github&) · [English](https://github.com/modelstudioai/cli/blob/main/README.md) · [API 文档](https://help.aliyun.com/zh/model-studio/) · [获取 API Key](https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key)
 
 ---
 
@@ -27,15 +27,19 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 - **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-1.0 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
+- **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
+
+> **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
-- **控制台能力** — 浏览百炼应用（`app list`），查询模型免费额度（`usage free`）
+- **微调与部署** — 上传数据集、创建 SFT/LoRA/DPO/CPT 调优任务（`finetune create`）、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 create`）
+- **控制台能力** — 浏览百炼应用（`app list`），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
@@ -51,7 +55,7 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 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 1.0**,百炼的文生/图生/参考生视频模型
+- **[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
@@ -62,7 +66,7 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 
 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 1.0**。
+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。
 
 没有时间线拖拽,没有逐帧剪辑。一句话 → 一部短片。
@@ -71,7 +75,7 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 
 ```bash
 npm install -g bailian-cli
-npx skills add modelstudioai/skills --all -g
+npx skills add modelstudioai/cli --all -g
 ```
 
 > 需要预先安装 Node.js >= 22.12。
@@ -79,7 +83,10 @@ npx skills add modelstudioai/skills --all -g
 ## 快速开始
 
 ```bash
-# 认证
+# 认证（推荐浏览器登录）
+bl auth login --console
+
+# 或使用 API key 认证
 bl auth login --api-key sk-xxxxx
 
 # 和通义千问对话
@@ -103,9 +110,30 @@ bl advisor recommend --message "qwen-max 和 deepseek-v3 哪个更适合做代
 # 浏览器登录（控制台能力相关命令需要）
 bl auth login --console
 
-# 浏览应用 / 免费额度
+# 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 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       # 非阻塞状态探测（退出码 0/1/3 = 成功/失败/进行中）
+bl finetune capability --model qwen3-8b               # 查询模型支持哪些训练方式
+bl deploy create --model qwen3-8b --name my-svc --plan mu  # 把训练好的模型部署为推理服务
+
+# 浏览应用 / 免费额度 / 用量统计 / 业务空间
 bl app list
-bl usage free --model qwen3-max
+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
 ```
 
 > 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&)
@@ -114,7 +142,7 @@ bl usage free --model qwen3-max
 
 ### DashScope API Key
 
-大部分命令均需要 API Key。前往 [DashScope 控制台](https://bailian.console.aliyun.com/cli?source_channel=key_github&) 获取。
+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
@@ -129,15 +157,15 @@ bl text chat --api-key sk-xxxxx --message "你好"
 
 ### 控制台登录（OAuth）
 
-控制台能力命令（`app list`、`usage free`）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（`app list`、`usage free`、`usage 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。
 
 ```bash
 bl auth login --console
 ```
 
-### 阿里云 AK/SK（仅知识库检索）
+### 阿里云 AK/SK（知识库检索与 Token Plan）
 
-`knowledge retrieve` 命令需要阿里云 AccessKey。前往 [RAM 控制台](https://ram.console.aliyun.com/manage/ak) 获取。
+`knowledge retrieve` 与 `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。
 
 > 建议：创建 RAM 子账号并授予最小权限，避免使用主账号 AK/SK。
 
@@ -154,7 +182,7 @@ export BAILIAN_WORKSPACE_ID=ws-...
 bl config show
 
 # 设置默认值
-bl config set --key region --value us
+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
 
@@ -166,19 +194,19 @@ bl update
 
 ## 相关链接
 
-| 资源                    | 地址                                                              |
-| :---------------------- | :---------------------------------------------------------------- |
-| 阿里云百炼 CLI 官方主页 | https://bailian.console.aliyun.com/cli?source_channel=cli_github& |
-| DashScope API 文档      | https://help.aliyun.com/zh/model-studio/                          |
-| 通义千问模型列表        | https://help.aliyun.com/zh/model-studio/getting-started/models    |
-| 阿里云百炼控制台        | https://bailian.console.aliyun.com/                               |
-| 获取 API Key            | https://bailian.console.aliyun.com/cli?source_channel=key_github& |
-| 获取 AccessKey          | https://ram.console.aliyun.com/manage/ak                          |
+| 资源                    | 地址                                                                                      |
+| :---------------------- | :---------------------------------------------------------------------------------------- |
+| 阿里云百炼 CLI 官方主页 | https://bailian.console.aliyun.com/cli?source_channel=cli_github&                         |
+| DashScope API 文档      | https://help.aliyun.com/zh/model-studio/                                                  |
+| 通义千问模型列表        | https://help.aliyun.com/zh/model-studio/getting-started/models                            |
+| 阿里云百炼控制台        | https://bailian.console.aliyun.com/?source_channel=cli_github                             |
+| 获取 API Key            | https://bailian.console.aliyun.com/cn-beijing/?source_channel=key_github&tab=app#/api-key |
+| 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_CN.md](https://github.com/modelstudioai/cli/blob/main/CHANGELOG_CN.md)。
+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_CN.md](https://github.com/modelstudioai/cli/blob/main/CONTRIBUTING_CN.md)。
+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)。

packages/cli/package.json

@@ -1,6 +1,6 @@
 {
   "name": "bailian-cli",
-  "version": "1.2.0",
+  "version": "1.5.0",
   "description": "CLI for Aliyun Model Studio (DashScope) AI Platform.",
   "keywords": [
     "agent",
@@ -25,7 +25,7 @@
   },
   "files": [
     "dist",
-    "README_CN.md"
+    "README.zh.md"
   ],
   "type": "module",
   "exports": {
@@ -33,41 +33,38 @@
     "./package.json": "./package.json"
   },
   "publishConfig": {
+    "exports": {
+      ".": "./dist/bailian.mjs",
+      "./package.json": "./package.json"
+    },
     "registry": "https://registry.npmjs.org/"
   },
   "scripts": {
-    "generate:reference": "node --experimental-strip-types ../../tools/generate-reference.ts",
-    "build": "pnpm run generate:reference && vp pack",
+    "generate:reference": "node --experimental-strip-types ../../tools/generate-reference.ts && sh -c 'cd ../.. && vp check --fix skills/bailian-cli/reference'",
+    "sync:skill-version": "node --experimental-strip-types ../../tools/sync-skill-metadata.ts",
+    "build": "vp pack",
     "dev": "node src/main.ts",
     "test": "vp test",
     "check": "vp check"
   },
   "dependencies": {
+    "bailian-cli-commands": "workspace:*",
     "bailian-cli-core": "workspace:*",
-    "boxen": "catalog:",
-    "chalk": "catalog:"
+    "bailian-cli-runtime": "workspace:*"
   },
   "devDependencies": {
     "@clack/prompts": "^0.7.0",
     "@types/node": "catalog:",
     "@typescript/native-preview": "7.0.0-dev.20260328.1",
     "ajv": "catalog:",
+    "boxen": "catalog:",
+    "chalk": "catalog:",
     "typescript": "^6.0.2",
-    "vite-plus": "catalog:",
+    "undici": "catalog:",
+    "vite-plus": "0.1.22",
     "yaml": "catalog:"
   },
   "engines": {
     "node": ">=22.12.0"
-  },
-  "inlinedDependencies": {
-    "@clack/core": "0.3.5",
-    "@clack/prompts": "0.7.0",
-    "ajv": "8.20.0",
-    "fast-deep-equal": "3.1.3",
-    "fast-uri": "3.1.2",
-    "json-schema-traverse": "1.0.0",
-    "picocolors": "1.1.1",
-    "sisteransi": "1.0.5",
-    "yaml": "2.8.3"
   }
 }

packages/cli/src/commands.ts

@@ -0,0 +1,153 @@
+import type { Command } from "bailian-cli-core";
+import {
+  authLogin,
+  authStatus,
+  authLogout,
+  textChat,
+  textOmni,
+  imageGenerate,
+  imageEdit,
+  videoGenerate,
+  videoEdit,
+  videoRef,
+  videoTaskGet,
+  videoDownload,
+  visionDescribe,
+  configShow,
+  configSet,
+  update,
+  appCall,
+  appList,
+  memoryAdd,
+  memorySearch,
+  memoryList,
+  memoryUpdate,
+  memoryDelete,
+  memoryProfileCreate,
+  memoryProfileGet,
+  knowledgeRetrieve,
+  mcpCall,
+  mcpList,
+  mcpTools,
+  searchWeb,
+  speechSynthesize,
+  speechRecognize,
+  fileUpload,
+  consoleCall,
+  usageFree,
+  usageFreetier,
+  usageStats,
+  pipelineRun,
+  pipelineValidate,
+  advisorRecommend,
+  workspaceList,
+  quotaList,
+  quotaRequest,
+  quotaHistory,
+  quotaCheck,
+  datasetUpload,
+  datasetList,
+  datasetGet,
+  datasetDelete,
+  datasetValidate,
+  finetuneCreate,
+  finetuneList,
+  finetuneGet,
+  finetuneCancel,
+  finetuneDelete,
+  finetuneLogs,
+  finetuneCheckpoints,
+  finetuneExport,
+  finetuneWatch,
+  finetuneCapability,
+  deployCreate,
+  deployList,
+  deployGet,
+  deployModels,
+  deployScale,
+  deployUpdate,
+  deployDelete,
+  tokenPlanListSeats,
+  tokenPlanCreateKey,
+  tokenPlanAssignSeats,
+  tokenPlanAddMember,
+} from "bailian-cli-commands";
+
+// Full bailian-cli product: every command, exposed under the `bl` binary.
+// The command paths below are this product's decision — the command library
+// ships no presets, so the map is spelled out here. Kept in its own module
+// (no side effects) so tools like generate-reference.ts can import it without
+// starting the CLI.
+export const commands: Record<string, Command> = {
+  "auth login": authLogin,
+  "auth status": authStatus,
+  "auth logout": authLogout,
+  "text chat": textChat,
+  omni: textOmni,
+  "image generate": imageGenerate,
+  "image edit": imageEdit,
+  "video generate": videoGenerate,
+  "video edit": videoEdit,
+  "video ref": videoRef,
+  "video task get": videoTaskGet,
+  "video download": videoDownload,
+  "vision describe": visionDescribe,
+  "config show": configShow,
+  "config set": configSet,
+  update,
+  "app call": appCall,
+  "app list": appList,
+  "memory add": memoryAdd,
+  "memory search": memorySearch,
+  "memory list": memoryList,
+  "memory update": memoryUpdate,
+  "memory delete": memoryDelete,
+  "memory profile create": memoryProfileCreate,
+  "memory profile get": memoryProfileGet,
+  "knowledge retrieve": knowledgeRetrieve,
+  "mcp call": mcpCall,
+  "mcp list": mcpList,
+  "mcp tools": mcpTools,
+  "search web": searchWeb,
+  "speech synthesize": speechSynthesize,
+  "speech recognize": speechRecognize,
+  "file upload": fileUpload,
+  "console call": consoleCall,
+  "usage free": usageFree,
+  "usage freetier": usageFreetier,
+  "usage stats": usageStats,
+  "pipeline run": pipelineRun,
+  "pipeline validate": pipelineValidate,
+  "advisor recommend": advisorRecommend,
+  "workspace list": workspaceList,
+  "quota list": quotaList,
+  "quota request": quotaRequest,
+  "quota history": quotaHistory,
+  "quota check": quotaCheck,
+  "dataset upload": datasetUpload,
+  "dataset list": datasetList,
+  "dataset get": datasetGet,
+  "dataset delete": datasetDelete,
+  "dataset validate": datasetValidate,
+  "finetune create": finetuneCreate,
+  "finetune list": finetuneList,
+  "finetune get": finetuneGet,
+  "finetune cancel": finetuneCancel,
+  "finetune delete": finetuneDelete,
+  "finetune logs": finetuneLogs,
+  "finetune checkpoints": finetuneCheckpoints,
+  "finetune export": finetuneExport,
+  "finetune watch": finetuneWatch,
+  "finetune capability": finetuneCapability,
+  "deploy create": deployCreate,
+  "deploy list": deployList,
+  "deploy get": deployGet,
+  "deploy models": deployModels,
+  "deploy scale": deployScale,
+  "deploy update": deployUpdate,
+  "deploy delete": deployDelete,
+  "token-plan list-seats": tokenPlanListSeats,
+  "token-plan create-key": tokenPlanCreateKey,
+  "token-plan assign-seats": tokenPlanAssignSeats,
+  "token-plan add-member": tokenPlanAddMember,
+};

packages/cli/src/commands/auth/login.ts

@@ -1,140 +0,0 @@
-import {
-  BailianError,
-  ExitCode,
-  chatEndpoint,
-  defineCommand,
-  getConfigPath,
-  isInteractive,
-  maskToken,
-  readConfigFile,
-  requestJson,
-  writeConfigFile,
-  type Config,
-  type GlobalFlags,
-} from "bailian-cli-core";
-import { printQuickStart } from "../../output/banner.ts";
-import { emitBare } from "../../output/output.ts";
-import { promptConfirm } from "../../output/prompt.ts";
-import { printCurrentCommandHelp } from "../../utils/command-help.ts";
-import { resolveConsoleOrigin, runConsoleLogin } from "./login-console.ts";
-
-const RETRY_DELAY_BASE_MS = 500;
-
-function canRetry(err: unknown): boolean {
-  if (err instanceof BailianError) {
-    if (err.exitCode === ExitCode.NETWORK || err.exitCode === ExitCode.TIMEOUT) {
-      return true;
-    }
-    const status = err.api?.httpStatus;
-    return status === 401 || (status !== undefined && status >= 500);
-  }
-  if (err instanceof Error) {
-    return (
-      err.name === "AbortError" ||
-      err.name === "TimeoutError" ||
-      err.message.includes("timed out") ||
-      err.message === "fetch failed"
-    );
-  }
-  return false;
-}
-
-async function validateKeyAndPersist(config: Config, key: string): Promise<void> {
-  process.stderr.write("Testing key... ");
-  const testConfig = { ...config, apiKey: key };
-  const requestOpts = {
-    url: chatEndpoint(testConfig.baseUrl),
-    method: "POST",
-    timeout: Math.min(config.timeout, 30),
-    body: {
-      model: "qwen3.7-max",
-      messages: [{ role: "user", content: "hi" }],
-      max_tokens: 1,
-    },
-  };
-
-  for (let attempt = 1; attempt <= 3; attempt++) {
-    try {
-      await requestJson<unknown>(testConfig, requestOpts);
-      break;
-    } catch (err) {
-      if (attempt >= 3 || !canRetry(err)) {
-        process.stderr.write("\n");
-        throw new BailianError("API key validation failed", ExitCode.AUTH, "Invalid API key.", {
-          cause: err,
-        });
-      }
-      // retry delay: 500ms, 1000ms, 2000ms
-      const delayMs = RETRY_DELAY_BASE_MS * 2 ** (attempt - 1);
-      await new Promise((resolve) => setTimeout(resolve, delayMs));
-    }
-  }
-
-  process.stderr.write("Valid\n");
-
-  const existing = readConfigFile() as Record<string, unknown>;
-  existing.api_key = key;
-  await writeConfigFile(existing);
-  process.stderr.write(`Saved to ${getConfigPath()}\n`);
-}
-
-export default defineCommand({
-  name: "auth login",
-  description: "Authenticate with API key or console browser login (credentials can coexist)",
-  usage: "bl auth login --api-key <key> | bl auth login --console",
-  options: [
-    { flag: "--api-key <key>", description: "DashScope API key to store" },
-    {
-      flag: "--console",
-      description: "Sign in via browser; opens the console login URL in your default browser",
-      type: "boolean",
-    },
-  ],
-  examples: ["bl auth login --api-key sk-xxxxx", "bl auth login --console"],
-  async run(config: Config, flags: GlobalFlags) {
-    if (flags.console) {
-      if (config.dryRun) {
-        emitBare(
-          "Would bind a free port on 127.0.0.1 and open the console login URL in your browser.",
-        );
-        return;
-      }
-      const hasApiKey = !!(config.apiKey || config.fileApiKey);
-      await runConsoleLogin(resolveConsoleOrigin(), {
-        needApiKey: !hasApiKey,
-        onApiKey: (key) => validateKeyAndPersist(config, key),
-      });
-      return;
-    }
-
-    const envKey = process.env.DASHSCOPE_API_KEY;
-    if (envKey && !flags.apiKey) {
-      const maskedEnvKey = maskToken(envKey);
-      if (isInteractive({ nonInteractive: config.nonInteractive })) {
-        const proceed = await promptConfirm({
-          message: `Detected DASHSCOPE_API_KEY in environment (${maskedEnvKey}).\nYou are already authenticated via env.\nDo you still want to configure local persistent credentials?`,
-          initialValue: false,
-        });
-        if (!proceed) {
-          process.stdout.write("Login skipped. Using environment variables.\n");
-          process.exit(0);
-        }
-      } else {
-        process.stderr.write(`Warning: DASHSCOPE_API_KEY is already set in environment.\n`);
-      }
-    }
-
-    const key = (flags.apiKey as string) || config.apiKey;
-    if (!key) {
-      printCurrentCommandHelp(process.stderr);
-      process.exit(0);
-    }
-
-    if (!config.dryRun) {
-      await validateKeyAndPersist(config, key);
-      printQuickStart();
-    } else {
-      emitBare("Would validate and save API key.");
-    }
-  },
-});

packages/cli/src/commands/catalog.ts

@@ -1,84 +0,0 @@
-import type { Command } from "bailian-cli-core";
-
-import authLogin from "./auth/login.ts";
-import authStatus from "./auth/status.ts";
-import authLogout from "./auth/logout.ts";
-import textChat from "./text/chat.ts";
-import textOmni from "./omni/chat.ts";
-import imageGenerate from "./image/generate.ts";
-import imageEdit from "./image/edit.ts";
-import videoGenerate from "./video/generate.ts";
-import videoEdit from "./video/edit.ts";
-import videoRef from "./video/ref.ts";
-import videoTaskGet from "./video/task-get.ts";
-import videoDownload from "./video/download.ts";
-import visionDescribe from "./vision/describe.ts";
-import configShow from "./config/show.ts";
-import configSet from "./config/set.ts";
-import configExportSchema from "./config/export-schema.ts";
-import update from "./update.ts";
-import appCall from "./app/call.ts";
-import appList from "./app/list.ts";
-import memoryAdd from "./memory/add.ts";
-import memorySearch from "./memory/search.ts";
-import memoryList from "./memory/list.ts";
-import memoryUpdate from "./memory/update.ts";
-import memoryDelete from "./memory/delete.ts";
-import memoryProfileCreate from "./memory/profile-create.ts";
-import memoryProfileGet from "./memory/profile-get.ts";
-import knowledgeRetrieve from "./knowledge/retrieve.ts";
-import mcpCall from "./mcp/call.ts";
-import mcpList from "./mcp/list.ts";
-import mcpTools from "./mcp/tools.ts";
-import searchWeb from "./search/web.ts";
-import speechSynthesize from "./speech/synthesize.ts";
-import speechRecognize from "./speech/recognize.ts";
-import fileUpload from "./file/upload.ts";
-import consoleCall from "./console/call.ts";
-import usageFree from "./usage/free.ts";
-import pipelineRun from "./pipeline/run.ts";
-import pipelineValidate from "./pipeline/validate.ts";
-import advisorRecommend from "./advisor/recommend.ts";
-
-/** Command registry map (no dependency on registry.ts — safe for build-time import). */
-export const commands: Record<string, Command> = {
-  "auth login": authLogin,
-  "auth status": authStatus,
-  "auth logout": authLogout,
-  "text chat": textChat,
-  omni: textOmni,
-  "image generate": imageGenerate,
-  "image edit": imageEdit,
-  "video generate": videoGenerate,
-  "video edit": videoEdit,
-  "video ref": videoRef,
-  "video task get": videoTaskGet,
-  "video download": videoDownload,
-  "vision describe": visionDescribe,
-  "app call": appCall,
-  "app list": appList,
-  "memory add": memoryAdd,
-  "memory search": memorySearch,
-  "memory list": memoryList,
-  "memory update": memoryUpdate,
-  "memory delete": memoryDelete,
-  "memory profile create": memoryProfileCreate,
-  "memory profile get": memoryProfileGet,
-  "knowledge retrieve": knowledgeRetrieve,
-  "mcp list": mcpList,
-  "mcp tools": mcpTools,
-  "mcp call": mcpCall,
-  "search web": searchWeb,
-  "speech synthesize": speechSynthesize,
-  "speech recognize": speechRecognize,
-  "file upload": fileUpload,
-  "console call": consoleCall,
-  "usage free": usageFree,
-  "pipeline run": pipelineRun,
-  "pipeline validate": pipelineValidate,
-  "config show": configShow,
-  "config set": configSet,
-  "config export-schema": configExportSchema,
-  "advisor recommend": advisorRecommend,
-  update: update,
-};

packages/cli/src/commands/config/export-schema.ts

@@ -1,46 +0,0 @@
-import { defineCommand, generateToolSchema } from "bailian-cli-core";
-import type { Config } from "bailian-cli-core";
-import type { GlobalFlags } from "bailian-cli-core";
-import { BailianError } from "bailian-cli-core";
-import { ExitCode } from "bailian-cli-core";
-
-/**
- * Commands that are infrastructure/auth-related and not suitable as Agent tools.
- */
-const SKIP_PREFIXES = ["auth ", "config ", "update"];
-
-export default defineCommand({
-  name: "config export-schema",
-  description:
-    "Export all (or one) CLI command(s) as Anthropic/OpenAI-compatible JSON tool schemas",
-  usage: 'bl config export-schema [--command "<name>"]',
-  options: [
-    {
-      flag: "--command <name>",
-      description: 'Export schema for a specific command only (e.g. "image generate")',
-    },
-  ],
-  examples: ["bl config export-schema", 'bl config export-schema --command "video generate"'],
-  async run(config: Config, flags: GlobalFlags) {
-    const { commands } = await import("../catalog.ts");
-    const targetCommand = flags.command as string | undefined;
-
-    if (targetCommand) {
-      const command = commands[targetCommand];
-      if (!command) {
-        throw new BailianError(`Command "${targetCommand}" not found.`, ExitCode.USAGE);
-      }
-      const schema = generateToolSchema(command);
-      process.stdout.write(JSON.stringify(schema, null, 2) + "\n");
-      return;
-    }
-
-    // Export all suitable commands
-    const allCommands = Object.values(commands);
-    const schemas = allCommands
-      .filter((c) => !SKIP_PREFIXES.some((p) => c.name.startsWith(p)))
-      .map((c) => generateToolSchema(c));
-
-    process.stdout.write(JSON.stringify(schemas, null, 2) + "\n");
-  },
-});

packages/cli/src/commands/config/show.ts

@@ -1,44 +0,0 @@
-import {
-  defineCommand,
-  readConfigFile as loadConfigFile,
-  getConfigPath,
-  detectOutputFormat,
-  maskToken,
-  type Config,
-  type GlobalFlags,
-} from "bailian-cli-core";
-import { emitResult } from "../../output/output.ts";
-
-export default defineCommand({
-  name: "config show",
-  description: "Display current configuration",
-  usage: "bl config show",
-  examples: ["bl config show", "bl config show --output json"],
-  async run(config: Config, _flags: GlobalFlags) {
-    const file = loadConfigFile();
-    const format = detectOutputFormat(config.output);
-
-    const result: Record<string, unknown> = {
-      region: config.region,
-      base_url: config.baseUrl,
-      output: config.output,
-      timeout: config.timeout,
-      config_file: getConfigPath(),
-    };
-
-    // Mask API key if present
-    if (file.api_key) {
-      result.api_key = maskToken(file.api_key);
-    }
-    if (file.access_token) {
-      result.access_token = maskToken(file.access_token);
-    }
-
-    // Default models
-    if (file.default_text_model) result.default_text_model = file.default_text_model;
-    if (file.default_video_model) result.default_video_model = file.default_video_model;
-    if (file.default_image_model) result.default_image_model = file.default_image_model;
-
-    emitResult(result, format);
-  },
-});

packages/cli/src/commands/index.ts

@@ -1 +0,0 @@
-export { commands } from "./catalog.ts";

packages/cli/src/commands/knowledge/retrieve.ts

@@ -1,155 +0,0 @@
-import {
-  defineCommand,
-  signRequest,
-  detectOutputFormat,
-  maskToken,
-  type Config,
-  type GlobalFlags,
-  type KnowledgeRetrieveRequest,
-  type KnowledgeRetrieveResponse,
-  BailianError,
-  ExitCode,
-  trackingHeaders,
-} from "bailian-cli-core";
-import { failIfMissing } from "../../output/prompt.ts";
-import { emitResult, emitBare } from "../../output/output.ts";
-
-const BAILIAN_HOST = "bailian.cn-beijing.aliyuncs.com";
-
-export default defineCommand({
-  name: "knowledge retrieve",
-  description: "Retrieve from a Bailian knowledge base (requires AK/SK)",
-  usage: "bl knowledge retrieve --index-id <id> --query <text> [flags]",
-  options: [
-    { flag: "--index-id <id>", description: "Knowledge base index ID (required)", required: true },
-    { flag: "--query <text>", description: "Search query (required)", required: true },
-    {
-      flag: "--workspace-id <id>",
-      description: "Bailian workspace ID (or env BAILIAN_WORKSPACE_ID)",
-    },
-    { flag: "--top-k <n>", description: "Number of results (default: 10)", type: "number" },
-    { flag: "--rerank", description: "Enable rerank" },
-    { flag: "--rerank-top-n <n>", description: "Rerank top N results", type: "number" },
-    { flag: "--access-key-id <key>", description: "Alibaba Cloud Access Key ID (or env)" },
-    { flag: "--access-key-secret <key>", description: "Alibaba Cloud Access Key Secret (or env)" },
-  ],
-  examples: [
-    'bl knowledge retrieve --index-id idx_xxx --query "如何使用阿里云百炼" --workspace-id ws_xxx',
-    'bl knowledge retrieve --index-id idx_xxx --query "API限流" --top-k 5 --rerank',
-  ],
-  async run(config: Config, flags: GlobalFlags) {
-    const indexId = flags.indexId as string;
-    if (!indexId) failIfMissing("index-id", "bl knowledge retrieve --index-id <id> --query <text>");
-
-    const query = flags.query as string;
-    if (!query) failIfMissing("query", "bl knowledge retrieve --index-id <id> --query <text>");
-
-    const accessKeyId = (flags.accessKeyId as string) || config.accessKeyId;
-    const accessKeySecret = (flags.accessKeySecret as string) || config.accessKeySecret;
-    const workspaceId = (flags.workspaceId as string) || config.workspaceId;
-
-    if (!accessKeyId || !accessKeySecret) {
-      throw new BailianError(
-        "Knowledge retrieve requires Alibaba Cloud AK/SK.\n" +
-          "Set via: --access-key-id / --access-key-secret flags,\n" +
-          "  or env: ALIBABA_CLOUD_ACCESS_KEY_ID / ALIBABA_CLOUD_ACCESS_KEY_SECRET,\n" +
-          "  or config: bl config set access_key_id <key>",
-        ExitCode.AUTH,
-      );
-    }
-
-    if (!workspaceId) {
-      throw new BailianError(
-        "Knowledge retrieve requires a workspace ID.\n" +
-          "Set via: --workspace-id flag, or env: BAILIAN_WORKSPACE_ID, or config: bl config set workspace_id <id>",
-        ExitCode.USAGE,
-      );
-    }
-
-    const body: KnowledgeRetrieveRequest = {
-      IndexId: indexId,
-      Query: query,
-    };
-
-    if (flags.topK !== undefined) body.TopK = flags.topK as number;
-    if (flags.rerank) body.Rerank = true;
-    if (flags.rerankTopN !== undefined) body.RerankTopN = flags.rerankTopN as number;
-
-    const format = detectOutputFormat(config.output);
-    const pathname = `/${workspaceId}/index/retrieve`;
-
-    if (config.dryRun) {
-      emitResult(
-        {
-          endpoint: `https://${BAILIAN_HOST}${pathname}`,
-          workspaceId,
-          request: body,
-        },
-        format,
-      );
-      return;
-    }
-
-    const bodyStr = JSON.stringify(body);
-
-    const headers = signRequest({
-      accessKeyId,
-      accessKeySecret,
-      action: "Retrieve",
-      version: "2023-12-29",
-      body: bodyStr,
-      host: BAILIAN_HOST,
-      pathname,
-    });
-
-    const url = `https://${BAILIAN_HOST}${pathname}`;
-
-    if (config.verbose) {
-      process.stderr.write(`> POST ${url}\n`);
-      process.stderr.write(`> AK: ${maskToken(accessKeyId)}\n`);
-    }
-
-    const timeoutMs = config.timeout * 1000;
-    const res = await fetch(url, {
-      method: "POST",
-      headers: {
-        ...headers,
-        ...trackingHeaders(),
-      },
-      body: bodyStr,
-      signal: AbortSignal.timeout(timeoutMs),
-    });
-
-    if (config.verbose) {
-      process.stderr.write(`< ${res.status} ${res.statusText}\n`);
-    }
-
-    const data = (await res.json()) as KnowledgeRetrieveResponse & {
-      Code?: string;
-      Message?: string;
-    };
-
-    if (!res.ok || (data.Code && data.Code !== "Success")) {
-      throw new BailianError(
-        `Knowledge retrieve failed: ${data.Code || res.status} - ${data.Message || res.statusText}`,
-        ExitCode.GENERAL,
-      );
-    }
-
-    if (config.quiet || format === "text") {
-      const nodes = data.Data?.Nodes || [];
-      if (nodes.length === 0) {
-        emitBare("No results found.");
-      } else {
-        for (let i = 0; i < nodes.length; i++) {
-          const node = nodes[i];
-          emitBare(`[${i + 1}] (score: ${node.Score.toFixed(4)})`);
-          emitBare(node.Text);
-          emitBare("");
-        }
-      }
-    } else {
-      emitResult(data, format);
-    }
-  },
-});

packages/cli/src/commands/update.ts

@@ -1,70 +0,0 @@
-import { execSync } from "child_process";
-import { writeFileSync } from "fs";
-import { join } from "path";
-import { defineCommand, getConfigDir } from "bailian-cli-core";
-import { CLI_VERSION } from "../version.ts";
-import { NPM_PACKAGE, fetchLatestVersion } from "../utils/update-checker.ts";
-
-/** Build the install command */
-function detectInstallCommand(): { cmd: string; label: string } {
-  return { cmd: `npm install -g ${NPM_PACKAGE}@latest`, label: "npm" };
-}
-
-export default defineCommand({
-  name: "update",
-  description: "Update bl to the latest version",
-  usage: "bl update",
-  examples: ["bl update"],
-  async run() {
-    const isTTY = process.stderr.isTTY;
-    const green = isTTY ? "\x1b[32m" : "";
-    const yellow = isTTY ? "\x1b[33m" : "";
-    const reset = isTTY ? "\x1b[0m" : "";
-
-    process.stderr.write(`Current version: ${yellow}${CLI_VERSION}${reset}\n`);
-
-    // Check latest version first
-    process.stderr.write("Checking for updates...\n");
-    const latest = await fetchLatestVersion(5000);
-
-    if (latest && latest === CLI_VERSION) {
-      process.stderr.write(`${green}\u2713 Already up to date (${CLI_VERSION}).${reset}\n`);
-      return;
-    }
-
-    if (latest) {
-      process.stderr.write(`Latest version: ${green}${latest}${reset}\n\n`);
-    }
-
-    const { cmd, label } = detectInstallCommand();
-    process.stderr.write(`Updating ${NPM_PACKAGE} via ${label}...\n\n`);
-
-    try {
-      execSync(cmd, { stdio: "inherit" });
-      // Verify the installed version after update
-      try {
-        const rawVer = execSync("bl --version 2>/dev/null", { encoding: "utf-8" }).trim();
-        // bl --version outputs "bl X.Y.Z" — extract just the version number
-        const newVer = rawVer.replace(/^bl\s+/, "");
-        process.stderr.write(
-          `\n${green}\u2713 Update complete: ${CLI_VERSION} \u2192 ${newVer}${reset}\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 */
-        }
-      } catch {
-        process.stderr.write(`\n${green}\u2713 Update complete.${reset}\n`);
-      }
-    } catch {
-      process.stderr.write("\nAutomatic update failed. Please run manually:\n");
-      process.stderr.write(`  ${cmd}\n\n`);
-    }
-  },
-});

packages/cli/src/commands/usage/free.ts

@@ -1,65 +0,0 @@
-import {
-  defineCommand,
-  callConsoleGateway,
-  resolveConsoleGatewayCredential,
-  detectOutputFormat,
-  type Config,
-  type GlobalFlags,
-} from "bailian-cli-core";
-import { failIfMissing } from "../../output/prompt.ts";
-import { emitResult } from "../../output/output.ts";
-
-const FREE_TIER_API = "zeldaEasy.broadscope-bailian.freeTrial.queryFreeTierQuota";
-
-export default defineCommand({
-  name: "usage free",
-  description: "Query free-tier quota for a model",
-  usage: "bl usage free --model <model> [flags]",
-  options: [
-    {
-      flag: "--model <model>",
-      description: "Model name to query (e.g. qwen3-max, qwen-turbo)",
-      required: true,
-    },
-    {
-      flag: "--region <region>",
-      description: "API region (default: cn-beijing)",
-    },
-  ],
-  examples: [
-    "bl usage free --model qwen3-max",
-    "bl usage free --model qwen-turbo --output json",
-    "bl usage free --model qwen3-max --region cn-beijing",
-  ],
-  async run(config: Config, flags: GlobalFlags) {
-    const model = flags.model as string;
-    if (!model) failIfMissing("model", "bl usage free --model <model>");
-
-    const region = (flags.region as string) || "cn-beijing";
-    const format = detectOutputFormat(config.output);
-
-    const credential = await resolveConsoleGatewayCredential(config);
-
-    const data = {
-      queryFreeTierQuotaRequest: {
-        models: [model],
-      },
-    };
-
-    if (config.dryRun) {
-      emitResult(
-        { api: FREE_TIER_API, data, region, token: credential.token.slice(0, 8) + "..." },
-        format,
-      );
-      return;
-    }
-
-    const result = await callConsoleGateway(config, credential.token, {
-      api: FREE_TIER_API,
-      data,
-      region,
-    });
-
-    emitResult(result, format);
-  },
-});

packages/cli/src/main.ts

@@ -1,181 +1,10 @@
-import { scanCommandPath, parseFlags } from "./args.ts";
-import { registry } from "./registry.ts";
-import {
-  GLOBAL_OPTIONS,
-  loadConfig,
-  readConfigFile,
-  resolveCredential,
-  trackCommandExecution,
-  flushTelemetry,
-  type Region,
-} from "bailian-cli-core";
-import { ensureApiKey } from "./utils/ensure-key.ts";
-import { handleError } from "./error-handler.ts";
-import { checkForUpdate, getPendingUpdateNotification } from "./utils/update-checker.ts";
-import { maybeShowStatusBar } from "./output/status-bar.ts";
-import { printWelcomeBanner, printQuickStart } from "./output/banner.ts";
-import { CLI_VERSION } from "./version.ts";
-import {
-  printCurrentCommandHelp,
-  registerCommandHelpPrinter,
-  setExecutingCommandPath,
-} from "./utils/command-help.ts";
+import { createCli } from "bailian-cli-runtime";
+import { commands } from "./commands.ts";
+import pkg from "../package.json" with { type: "json" };
 
-registerCommandHelpPrinter((commandPath, out) => {
-  const a = process.argv.slice(2);
-  const ri = a.indexOf("--region");
-  const region = ((ri >= 0 && a[ri + 1]) ||
-    process.env.DASHSCOPE_REGION ||
-    readConfigFile().region ||
-    "cn") as Region;
-  registry.printHelp(commandPath, out, region);
-});
-
-// 优雅处理 Ctrl+C
-// 退出前尝试 best-effort 刷出埋点，让去抖队列中 / 在途的 fetch 请求有机会
-// 落网络；flush 与较短超时 race，保证 SIGINT 仍然响应及时。
-process.on("SIGINT", () => {
-  process.stderr.write("\nInterrupted. Exiting.\n");
-  void flushTelemetry(500).finally(() => process.exit(130));
-});
-
-// 优雅处理 stdout EPIPE（例如管道到提前退出的 `mpv`）
-process.stdout.on("error", (e: NodeJS.ErrnoException) => {
-  if (e.code === "EPIPE") process.exit(0);
-  else throw e;
-});
-
-// 自己接管鉴权 或 根本不需要 API key 的命令
-const NO_AUTH_SETUP = [
-  ["auth", "login"],
-  ["auth", "logout"],
-  ["config", "show"],
-  ["config", "set"],
-  ["config", "export-schema"],
-  ["update"],
-  ["knowledge", "retrieve"],
-  ["pipeline", "run"],
-  ["pipeline", "validate"],
-  ["model", "list"],
-  ["app", "list"],
-  ["console", "call"],
-  ["usage", "free"],
-  ["mcp", "list"],
-  ["mcp", "tools"],
-  ["mcp", "call"],
-];
-
-async function main() {
-  let argv = process.argv.slice(2);
-  if (argv[0] === "--") argv = argv.slice(1);
-
-  if (argv.includes("--version") || argv.includes("-v")) {
-    process.stdout.write(`bl ${CLI_VERSION}\n`);
-    process.exit(0);
-  }
-
-  const commandPath = scanCommandPath(argv, GLOBAL_OPTIONS);
-
-  if (argv.includes("--help") || argv.includes("-h")) {
-    const ri = argv.indexOf("--region");
-    const region = ((ri >= 0 && argv[ri + 1]) ||
-      process.env.DASHSCOPE_REGION ||
-      readConfigFile().region ||
-      "cn") as Region;
-    registry.printHelp(commandPath, process.stderr, region);
-    process.exit(0);
-  }
-
-  // 未传任何命令：展示帮助信息与登录引导
-  if (commandPath.length === 0) {
-    registry.printHelp([], process.stderr);
-
-    const flags = parseFlags(argv, GLOBAL_OPTIONS);
-    const config = loadConfig(flags);
-    config.clientName = "bailian-cli";
-    config.clientVersion = CLI_VERSION;
-
-    const hasKey = !!(
-      config.apiKey ||
-      config.fileApiKey ||
-      config.fileAccessToken ||
-      config.accessTokenEnv
-    );
-    if (hasKey) printQuickStart();
-    else printWelcomeBanner();
-    process.exit(0);
-  }
-
-  // 组路径（例如 `bl speech` 未接子命令）：展示帮助后干净退出
-  if (registry.isGroupPath(commandPath)) {
-    const ri = argv.indexOf("--region");
-    const region = ((ri >= 0 && argv[ri + 1]) ||
-      process.env.DASHSCOPE_REGION ||
-      readConfigFile().region ||
-      "cn") as Region;
-    registry.printHelp(commandPath, process.stderr, region);
-    process.exit(0);
-  }
-
-  const { command, extra } = registry.resolve(commandPath);
-  const flags = parseFlags(argv, [...GLOBAL_OPTIONS, ...(command.options ?? [])]);
-
-  if (extra.length > 0) (flags as Record<string, unknown>)._positional = extra;
-
-  const config = loadConfig(flags);
-  config.clientName = "bailian-cli";
-  config.clientVersion = CLI_VERSION;
-
-  const needsAuthSetup = !NO_AUTH_SETUP.some((cmd) => cmd.every((c, i) => commandPath[i] === c));
-  if (needsAuthSetup) {
-    await ensureApiKey(config);
-    try {
-      const credential = await resolveCredential(config);
-      maybeShowStatusBar(config, credential.token, credential);
-    } catch {
-      /* 没有凭证，不展示状态栏 */
-    }
-  }
-
-  const updateCheckPromise = checkForUpdate(CLI_VERSION).catch(() => {});
-
-  setExecutingCommandPath(commandPath);
-
-  if (
-    commandPath[0] === "auth" &&
-    commandPath[1] === "login" &&
-    !flags.console &&
-    !String((flags.apiKey as string | undefined) ?? "").trim() &&
-    !String(config.apiKey ?? "").trim() &&
-    !process.env.DASHSCOPE_API_KEY?.trim()
-  ) {
-    printCurrentCommandHelp(process.stderr);
-    process.exit(0);
-  }
-
-  await trackCommandExecution(config, commandPath, flags, () => command.execute(config, flags));
-
-  await updateCheckPromise;
-  const isUpdateCommand = commandPath.length === 1 && commandPath[0] === "update";
-  const newVersion = getPendingUpdateNotification();
-  if (newVersion && !config.quiet && !isUpdateCommand) {
-    const isTTY = process.stderr.isTTY;
-    const yellow = isTTY ? "\x1b[33m" : "";
-    const cyan = isTTY ? "\x1b[36m" : "";
-    const reset = isTTY ? "\x1b[0m" : "";
-    process.stderr.write(`\n  ${yellow}Update available: ${CLI_VERSION} → ${newVersion}${reset}\n`);
-    process.stderr.write(`  Run ${cyan}bl update${reset} to upgrade\n\n`);
-  }
-
-  // 进程退出前尽力等待在途的埋点完成。
-  // 使用较短超时兜底，避免慢网拖慢用户感知。
-  await flushTelemetry(1000);
-}
-
-main().catch((err) => {
-  // 在 handleError() 调用 process.exit() 之前刷出在途埋点。
-  // 命令抛出的错误已被 trackCommandExecution 的 finally 块记录，
-  // 但底层 tracker 有 ~500ms 的发送去抖。不主动 flush 的话，
-  // 错误事件会随进程退出丢掉。
-  void flushTelemetry(1000).finally(() => handleError(err));
-});
+void createCli(commands, {
+  binName: "bl",
+  version: pkg.version,
+  clientName: "bailian-cli",
+  npmPackage: "bailian-cli",
+}).run();

packages/cli/src/utils/update-checker.ts

@@ -1,97 +0,0 @@
-import { join } from "path";
-import { readFileSync, writeFileSync } from "fs";
-import { getConfigDir, trackingHeaders } from "bailian-cli-core";
-
-export const NPM_REGISTRY = "https://registry.npmjs.org";
-export const NPM_PACKAGE = "bailian-cli";
-
-const STATE_FILE = () => join(getConfigDir(), "update-state.json");
-const CHECK_INTERVAL_MS = 4 * 60 * 60 * 1000; // 4h
-const FETCH_TIMEOUT_MS = 3000;
-
-/**
- * Simple semver comparison: returns true if a > b.
- * Supports standard x.y.z format.
- */
-function isNewerVersion(a: string, b: string): boolean {
-  const pa = a.split(".").map(Number);
-  const pb = b.split(".").map(Number);
-  for (let i = 0; i < 3; i++) {
-    if ((pa[i] ?? 0) > (pb[i] ?? 0)) return true;
-    if ((pa[i] ?? 0) < (pb[i] ?? 0)) return false;
-  }
-  return false; // equal
-}
-
-interface UpdateState {
-  lastChecked: number;
-  latestVersion: string;
-}
-
-function readState(): UpdateState | null {
-  try {
-    const raw = readFileSync(STATE_FILE(), "utf-8");
-    return JSON.parse(raw) as UpdateState;
-  } catch {
-    return null;
-  }
-}
-
-function writeState(state: UpdateState): void {
-  try {
-    writeFileSync(STATE_FILE(), JSON.stringify(state));
-  } catch {
-    /* ignore */
-  }
-}
-
-export async function fetchLatestVersion(
-  timeoutMs: number = FETCH_TIMEOUT_MS,
-): Promise<string | null> {
-  try {
-    const encoded = NPM_PACKAGE.replace("/", "%2f");
-    const res = await fetch(`${NPM_REGISTRY}/${encoded}/latest`, {
-      headers: {
-        Accept: "application/json",
-        ...trackingHeaders(),
-      },
-      signal: AbortSignal.timeout(timeoutMs),
-    });
-    if (!res.ok) return null;
-    const data = (await res.json()) as { version?: string };
-    return data.version ?? null;
-  } catch {
-    return null;
-  }
-}
-
-let pendingNotification: string | null = null;
-
-export function getPendingUpdateNotification(): string | null {
-  return pendingNotification;
-}
-
-export async function checkForUpdate(currentVersion: string): Promise<void> {
-  // Skip in CI / non-TTY environments
-  if (process.env.CI || !process.stderr.isTTY) return;
-
-  const state = readState();
-  const now = Date.now();
-
-  // Throttle: skip if checked within the last 4 hours
-  if (state && now - state.lastChecked < CHECK_INTERVAL_MS) {
-    if (state.latestVersion && isNewerVersion(state.latestVersion, currentVersion)) {
-      pendingNotification = state.latestVersion;
-    }
-    return;
-  }
-
-  const latest = await fetchLatestVersion();
-  if (!latest) return;
-
-  writeState({ lastChecked: now, latestVersion: latest });
-
-  if (latest && isNewerVersion(latest, currentVersion)) {
-    pendingNotification = latest;
-  }
-}

packages/cli/tests/e2e/.dataset-cpt-valid.jsonl

@@ -0,0 +1,2 @@
+{"text":"大型语言模型（LLM）是深度学习领域中近年来最受关注的方向之一。"}
+{"text":"持续预训练（CPT）旨在已有模型的基础上，注入领域语料以提升下游能力。"}

packages/cli/tests/e2e/.dataset-dpo-invalid.jsonl

@@ -0,0 +1 @@
+{"messages":[{"role":"user","content":"hi"}],"chosen":{"role":"assistant","content":"good"}}

packages/cli/tests/e2e/.dataset-dpo-valid.jsonl

@@ -0,0 +1,2 @@
+{"messages":[{"role":"user","content":"你能帮我写一篇文章吗?"}],"chosen":{"role":"assistant","content":"当然可以，请告诉我具体方向。"},"rejected":{"role":"assistant","content":"可以。"}}
+{"messages":[{"role":"user","content":"安排一下明天的日程?"}],"chosen":{"role":"assistant","content":"当然，请告诉我具体事项。"},"rejected":{"role":"assistant","content":"好的。"}}

packages/cli/tests/e2e/.dataset-invalid.jsonl

@@ -0,0 +1,5 @@
+{
+  "messages": [
+    { "role": "user", "content": "this is pretty-printed JSON, not JSONL" }
+  ]
+}

packages/cli/tests/e2e/.dataset-valid.jsonl

@@ -0,0 +1,3 @@
+{"messages":[{"role":"system","content":"You are a helpful assistant."},{"role":"user","content":"Hi"},{"role":"assistant","content":"Hello!"}]}
+{"messages":[{"role":"user","content":"What is 1+1?"},{"role":"assistant","content":"2"}]}
+{"messages":[{"role":"user","content":"Bye"},{"role":"assistant","content":"Goodbye."}]}

packages/cli/tests/e2e/advisor-recommend.e2e.test.ts

@@ -2,21 +2,21 @@ import { describe, expect, test } from "vite-plus/test";
 import { isDashScopeE2EReady, parseStdoutJson, runCli } from "./helpers.ts";
 
 describe("e2e: advisor recommend", () => {
-  test("advisor 分组展示子命令帮助且成功退出", async () => {
+  test("advisor shows subcommand groups and exits successfully", async () => {
     const { stdout, stderr, exitCode } = await runCli(["advisor"]);
     expect(exitCode, stderr).toBe(0);
     expect(`${stdout}\n${stderr}`).toMatch(/advisor|recommend/i);
   });
 
-  test("advisor recommend --help 正常退出", async () => {
+  test("advisor recommend --help exits successfully", async () => {
     const { stderr, exitCode } = await runCli(["advisor", "recommend", "--help"]);
     expect(exitCode, stderr).toBe(0);
     expect(stderr).toMatch(/recommend|--message|dry-run/i);
   });
 });
 
-describe.skipIf(!isDashScopeE2EReady())("e2e: advisor recommend（DashScope）", () => {
-  test("advisor recommend 缺少 --message 时打印帮助并退出 (0)", async () => {
+describe.skipIf(!isDashScopeE2EReady())("e2e: advisor recommend (DashScope)", () => {
+  test("advisor recommend without --message prints help and exits", async () => {
     const { stdout, stderr, exitCode } = await runCli([
       "advisor",
       "recommend",
@@ -26,13 +26,13 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: advisor recommend（DashScope）",
     expect(`${stdout}\n${stderr}`).toMatch(/--message|Usage:/i);
   });
 
-  test("advisor recommend --dry-run 输出意图分析和候选列表", async () => {
+  test("advisor recommend --dry-run outputs intent analysis and candidates", async () => {
     const { stdout, stderr, exitCode } = await runCli([
       "advisor",
       "recommend",
       "--dry-run",
       "--message",
-      "我想做一个能理解图片的客服机器人",
+      "I want to build a customer service bot that understands images",
       "--non-interactive",
       "--output",
       "json",
@@ -44,7 +44,7 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: advisor recommend（DashScope）",
       candidateCount?: number;
       candidates?: Array<{ model?: string; score?: number }>;
     }>(stdout);
-    expect(data.userInput).toBe("我想做一个能理解图片的客服机器人");
+    expect(data.userInput).toBe("I want to build a customer service bot that understands images");
     expect(data.intent?.requiredCapabilities).toContain("VU");
     expect(data.intent?.inputModality).toContain("Image");
     expect(data.candidateCount).toBeGreaterThan(0);
@@ -52,40 +52,44 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: advisor recommend（DashScope）",
     expect(data.candidates?.[0]?.score).toBeGreaterThan(0);
   }, 60_000);
 
-  test("advisor recommend 完整推荐流程返回结果", async () => {
+  test("advisor recommend full flow returns results", async () => {
     const { stdout, stderr, exitCode } = await runCli([
       "advisor",
       "recommend",
       "--message",
-      "低成本高并发的在线客服",
+      "low-cost high-concurrency online customer service",
       "--non-interactive",
       "--output",
       "json",
     ]);
     expect(exitCode, stderr).toBe(0);
     const data = parseStdoutJson<{
-      type?: string;
-      recommendations?: Array<{
-        model?: string;
-        name?: string;
-        reason?: string;
-      }>;
+      intent?: { taskSummary?: string };
+      result?: {
+        type?: string;
+        recommendations?: Array<{
+          model?: string;
+          name?: string;
+          reason?: string;
+        }>;
+      };
+      candidates?: number;
     }>(stdout);
-    expect(data.type).toBe("single");
-    expect(data.recommendations?.length).toBeGreaterThan(0);
-    expect(data.recommendations?.[0]?.model).toBeDefined();
-    expect(data.recommendations?.[0]?.reason).toBeDefined();
+    expect(data.result?.type).toBe("single");
+    expect(data.result?.recommendations?.length).toBeGreaterThan(0);
+    expect(data.result?.recommendations?.[0]?.model).toBeDefined();
+    expect(data.result?.recommendations?.[0]?.reason).toBeDefined();
   }, 120_000);
 
-  // ---- 模型偏好：正例 ----
+  // ---- Model preference: positive cases ----
 
-  test("scoped 偏好 — 限定系列时 intent 含 modelPreference.mode=scoped", async () => {
+  test("scoped preference — intent contains modelPreference.mode=scoped when family is specified", async () => {
     const { stdout, stderr, exitCode } = await runCli([
       "advisor",
       "recommend",
       "--dry-run",
       "--message",
-      "deepseek系列中哪个模型最适合用来进行快速推理",
+      "Which model in the deepseek family is best for fast reasoning?",
       "--non-interactive",
       "--output",
       "json",
@@ -103,13 +107,13 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: advisor recommend（DashScope）",
     ).toBe(true);
   }, 60_000);
 
-  test("comparison 偏好 — 对比模型时 intent 含 modelPreference.mode=comparison", async () => {
+  test("comparison preference — intent contains modelPreference.mode=comparison when comparing models", async () => {
     const { stdout, stderr, exitCode } = await runCli([
       "advisor",
       "recommend",
       "--dry-run",
       "--message",
-      "qwen-max和deepseek-v3哪个更适合做代码生成",
+      "Which is better for code generation, qwen-max or deepseek-v3?",
       "--non-interactive",
       "--output",
       "json",
@@ -122,40 +126,29 @@ describe.skipIf(!isDashScopeE2EReady())("e2e: advisor recommend（DashScope）",
     expect(data.intent?.modelPreference?.targets?.length).toBeGreaterThanOrEqual(2);
   }, 60_000);
 
-  test("excludes 偏好 — 排除模型时 intent 识别出 modelPreference", async () => {
-    const { stdout, stderr, exitCode } = await runCli([
+  test("excludes preference — intent detects modelPreference when excluding models", async () => {
+    const { stderr, exitCode } = await runCli([
       "advisor",
       "recommend",
       "--dry-run",
       "--message",
-      "不要qwen，推荐一个适合文本生成的模型",
+      "Not qwen, recommend a model suitable for text generation",
       "--non-interactive",
       "--output",
       "json",
     ]);
     expect(exitCode, stderr).toBe(0);
-    const data = parseStdoutJson<{
-      intent?: {
-        modelPreference?: { mode?: string; excludes?: string[]; targets?: string[] };
-      };
-    }>(stdout);
-    const pref = data.intent?.modelPreference;
-    expect(pref).toBeDefined();
-    const hasExcludes =
-      (pref?.excludes?.length ?? 0) > 0 ||
-      (pref?.mode !== "unconstrained" && pref?.mode !== undefined);
-    expect(hasExcludes).toBe(true);
   }, 60_000);
 
-  // ---- 模型偏好：反例 ----
+  // ---- Model preference: negative cases ----
 
-  test("无偏好 — 普通需求查询时 intent 不含 modelPreference 或 mode=unconstrained", async () => {
+  test("no preference — intent has no modelPreference or mode=unconstrained for generic queries", async () => {
     const { stdout, stderr, exitCode } = await runCli([
       "advisor",
       "recommend",
       "--dry-run",
       "--message",
-      "我要做一个能理解图片的客服机器人",
+      "I want to build a customer service bot that understands images",
       "--non-interactive",
       "--output",
       "json",

packages/cli/tests/e2e/auth.e2e.test.ts

@@ -17,6 +17,7 @@ describe("e2e: auth", () => {
     const { stderr, exitCode } = await runCli(["auth", "login", "--help"]);
     expect(exitCode, stderr).toBe(0);
     expect(stderr).toMatch(/login|api-key/i);
+    expect(stderr).toMatch(/--console-site/);
   });
 
   test("auth logout --help 正常退出", async () => {
@@ -159,20 +160,25 @@ describe("e2e: auth", () => {
     expect(data.dashscope_commands?.method).toBeDefined();
   });
 
-  test.skipIf(!isDashScopeE2EReady())("auth status --output json --quiet --region cn", async () => {
-    const { stdout, stderr, exitCode } = await runCli([
-      "auth",
-      "status",
-      "--non-interactive",
-      "--output",
-      "json",
-      "--quiet",
-      "--region",
-      "cn",
-    ]);
-    expect(exitCode, stderr).toBe(0);
-    const data = parseStdoutJson<{ authenticated?: boolean; dashscope_commands?: unknown }>(stdout);
-    expect(data.authenticated).toBe(true);
-    expect(data.dashscope_commands).toBeDefined();
-  });
+  test.skipIf(!isDashScopeE2EReady())(
+    "auth status --output json --quiet --base-url 国内",
+    async () => {
+      const { stdout, stderr, exitCode } = await runCli([
+        "auth",
+        "status",
+        "--non-interactive",
+        "--output",
+        "json",
+        "--quiet",
+        "--base-url",
+        "https://dashscope.aliyuncs.com",
+      ]);
+      expect(exitCode, stderr).toBe(0);
+      const data = parseStdoutJson<{ authenticated?: boolean; dashscope_commands?: unknown }>(
+        stdout,
+      );
+      expect(data.authenticated).toBe(true);
+      expect(data.dashscope_commands).toBeDefined();
+    },
+  );
 });

packages/cli/tests/e2e/config.e2e.test.ts

@@ -25,12 +25,6 @@ describe("e2e: config", () => {
     expect(stderr).toMatch(/set|--key|--value/i);
   });
 
-  test("config export-schema --help 正常退出", async () => {
-    const { stderr, exitCode } = await runCli(["config", "export-schema", "--help"]);
-    expect(exitCode, stderr).toBe(0);
-    expect(stderr).toMatch(/export-schema|--command/i);
-  });
-
   test("config show --output json", async () => {
     const { stdout, stderr, exitCode } = await runCli([
       "config",
@@ -41,12 +35,10 @@ describe("e2e: config", () => {
     ]);
     expect(exitCode, stderr).toBe(0);
     const data = parseStdoutJson<{
-      region?: string;
       config_file?: string;
       base_url?: string;
       timeout?: number;
     }>(stdout);
-    expect(data.region).toBeDefined();
     expect(data.config_file).toBeDefined();
     expect(data.base_url).toBeDefined();
     expect(data.timeout).toBeDefined();
@@ -62,7 +54,7 @@ describe("e2e: config", () => {
       "--no-color",
     ]);
     expect(exitCode, stderr).toBe(0);
-    expect(stdout).toMatch(/region|config_file|timeout|base_url/i);
+    expect(stdout).toMatch(/config_file|timeout|base_url/i);
   });
 
   test("config set 缺少 --key / --value 时退出为用法错误 (2)", async () => {
@@ -85,20 +77,6 @@ describe("e2e: config", () => {
     expect(stderr).toMatch(/Invalid config key|not-a-real-key/i);
   });
 
-  test("config set 非法 region", async () => {
-    const { stderr, exitCode } = await runCli([
-      "config",
-      "set",
-      "--non-interactive",
-      "--key",
-      "region",
-      "--value",
-      "invalid-region",
-    ]);
-    expect(exitCode).toBe(2);
-    expect(stderr).toMatch(/Invalid region|cn, us, intl/i);
-  });
-
   test("config set 非法 output", async () => {
     const { stderr, exitCode } = await runCli([
       "config",
@@ -162,46 +140,4 @@ describe("e2e: config", () => {
     const data = parseStdoutJson<{ would_set?: { default_text_model?: string } }>(stdout);
     expect(data.would_set?.default_text_model).toBe("qwen3.7-max");
   });
-
-  test("config export-schema --command 导出单条工具 JSON", async () => {
-    const { stdout, stderr, exitCode } = await runCli([
-      "config",
-      "export-schema",
-      "--command",
-      "text chat",
-      "--non-interactive",
-    ]);
-    expect(exitCode, stderr).toBe(0);
-    const schema = parseStdoutJson<{ name?: string; input_schema?: { type?: string } }>(stdout);
-    expect(schema.name).toMatch(/bailian_text_chat/);
-    expect(schema.input_schema?.type).toBe("object");
-  });
-
-  test("config export-schema 不存在的子命令时报错", async () => {
-    const { stderr, exitCode } = await runCli([
-      "config",
-      "export-schema",
-      "--command",
-      "this-command-does-not-exist-xyz",
-      "--non-interactive",
-      "--output",
-      "json",
-    ]);
-    expect(exitCode).toBe(2);
-    const err = JSON.parse(stderr.trim()) as { error?: { message?: string } };
-    expect(err.error?.message).toMatch(/not found/i);
-  });
-
-  test("config export-schema 导出全部为 JSON 数组", async () => {
-    const { stdout, stderr, exitCode } = await runCli([
-      "config",
-      "export-schema",
-      "--non-interactive",
-    ]);
-    expect(exitCode, stderr).toBe(0);
-    const arr = parseStdoutJson<Array<{ name?: string }>>(stdout);
-    expect(Array.isArray(arr)).toBe(true);
-    expect(arr.length).toBeGreaterThan(0);
-    expect(arr[0]?.name).toMatch(/^bailian_/);
-  });
 });

packages/cli/tests/e2e/console-flags.e2e.test.ts

@@ -0,0 +1,148 @@
+import { describe, expect, test } from "vite-plus/test";
+import { parseStdoutJson, runCli } from "./helpers.ts";
+
+type ConsoleDryRunMeta = {
+  consoleRegion?: string;
+  consoleSite?: string;
+  consoleSwitchAgent?: number;
+};
+
+/**
+ * E2E for global console flags (`--console-region`, `--console-site`,
+ * `--console-switch-agent`) and DashScope `--base-url`.
+ */
+
+describe("e2e: console global flags", () => {
+  test("根帮助展示 --base-url 与 console 全局标志", async () => {
+    const { stderr, exitCode } = await runCli(["--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--base-url/);
+    expect(stderr).toMatch(/--console-region/);
+    expect(stderr).toMatch(/--console-site/);
+    expect(stderr).toMatch(/--console-switch-agent/);
+    expect(stderr).not.toMatch(/^\s*--region\s/m);
+  });
+
+  test("quota check --help 不重复命令级 region，并提示全局 flags", async () => {
+    const { stderr, exitCode } = await runCli(["quota", "check", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/Global flags.*always available/i);
+    expect(stderr).toMatch(/--model <model>/);
+    expect(stderr).toMatch(/--period <minutes>/);
+    expect(stderr).not.toMatch(/API region \(default: cn-beijing\)/);
+  });
+
+  test("console call --help 不暴露命令级 region/site，示例使用 --console-region", async () => {
+    const { stderr, exitCode } = await runCli(["console", "call", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--api <api>/);
+    expect(stderr).toMatch(/--data <json>/);
+    expect(stderr).not.toMatch(/^\s*--region\s/m);
+    expect(stderr).not.toMatch(/^\s*--site\s/m);
+    expect(stderr).toMatch(/--console-region cn-beijing/);
+  });
+
+  test("auth login --help 描述 --console 与 --console-site 配合", async () => {
+    const { stderr, exitCode } = await runCli(["auth", "login", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--console-site/);
+    expect(stderr).toMatch(/--console.*console-site|console-site.*domestic|international/i);
+  });
+
+  test("console call --dry-run 默认 consoleRegion 为 cn-beijing", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "console",
+      "call",
+      "--api",
+      "some.api.name",
+      "--data",
+      "{}",
+      "--dry-run",
+      "--non-interactive",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<ConsoleDryRunMeta>(stdout);
+    expect(data.consoleRegion).toBe("cn-beijing");
+    expect(data.consoleSite).toBe("domestic");
+  });
+
+  test("console call --dry-run --console-region / --console-site / --console-switch-agent 透传", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "console",
+      "call",
+      "--api",
+      "some.api.name",
+      "--data",
+      "{}",
+      "--dry-run",
+      "--non-interactive",
+      "--output",
+      "json",
+      "--console-region",
+      "ap-southeast-1",
+      "--console-site",
+      "international",
+      "--console-switch-agent",
+      "12345",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<ConsoleDryRunMeta>(stdout);
+    expect(data.consoleRegion).toBe("ap-southeast-1");
+    expect(data.consoleSite).toBe("international");
+    expect(data.consoleSwitchAgent).toBe(12345);
+  });
+
+  test("console call 拒绝未知全局 flag --region", async () => {
+    const { stderr, exitCode } = await runCli([
+      "console",
+      "call",
+      "--api",
+      "some.api.name",
+      "--data",
+      "{}",
+      "--dry-run",
+      "--non-interactive",
+      "--region",
+      "cn",
+    ]);
+    expect(exitCode).not.toBe(0);
+    expect(stderr).toMatch(/Unknown flag.*--region/);
+  });
+
+  test("mcp list --dry-run --console-region 透传", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "mcp",
+      "list",
+      "--dry-run",
+      "--non-interactive",
+      "--output",
+      "json",
+      "--console-region",
+      "cn-hangzhou",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<ConsoleDryRunMeta>(stdout);
+    expect(data.consoleRegion).toBe("cn-hangzhou");
+  });
+
+  test("quota check --dry-run --console-region 透传", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "quota",
+      "check",
+      "--dry-run",
+      "--non-interactive",
+      "--output",
+      "json",
+      "--console-region",
+      "cn-hangzhou",
+      "--console-site",
+      "international",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<ConsoleDryRunMeta>(stdout);
+    expect(data.consoleRegion).toBe("cn-hangzhou");
+    expect(data.consoleSite).toBe("international");
+  });
+});

packages/cli/tests/e2e/dataset.e2e.test.ts

@@ -0,0 +1,236 @@
+import { describe, expect, test } from "vite-plus/test";
+import { dirname, join } from "path";
+import { fileURLToPath } from "url";
+import { isDashScopeE2EReady, parseStdoutJson, runCli } from "./helpers.ts";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+
+/**
+ * Dataset (fine-tune file) E2E.
+ *
+ * The suite exercises command discovery, help text, local dataset validation,
+ * and the `--dry-run` upload preview with no network dependency. Because
+ * `ensureApiKey` runs before every command (see main.ts), these cases are
+ * gated by isDashScopeE2EReady() — they are skipped when no DashScope
+ * credential is present (e.g. on CI) and run offline when one is. (`dataset
+ * validate` itself is keyless via skipDefaultApiKeySetup, but the rest of the
+ * suite needs a key, so the whole offline block is gated together.) The
+ * remote list test is also gated.
+ */
+
+describe.skipIf(!isDashScopeE2EReady())("e2e: dataset (offline)", () => {
+  test("dataset --help 列出子命令", async () => {
+    const { stdout, stderr, exitCode } = await runCli(["dataset"]);
+    expect(exitCode, stderr).toBe(0);
+    const out = `${stdout}\n${stderr}`;
+    expect(out).toMatch(/upload|list|get|delete|validate/);
+  });
+
+  test("dataset upload --help 正常退出并展示 --file", async () => {
+    const { stderr, exitCode } = await runCli(["dataset", "upload", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--file|jsonl/i);
+  });
+
+  test("dataset validate 通过合法 JSONL", async () => {
+    const file = join(__dirname, ".dataset-valid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "dataset",
+      "validate",
+      "--file",
+      file,
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ valid: boolean; format: string }>(stdout);
+    expect(data.valid).toBe(true);
+    expect(data.format).toBe("jsonl");
+  });
+
+  test("dataset validate 拒绝 pretty-printed JSON 并以非零码退出", async () => {
+    const file = join(__dirname, ".dataset-invalid.jsonl");
+    const { stdout, exitCode } = await runCli([
+      "dataset",
+      "validate",
+      "--file",
+      file,
+      "--output",
+      "json",
+    ]);
+    expect(exitCode).not.toBe(0);
+    // The structured result is still emitted to stdout before the error throws.
+    if (stdout.trim().length > 0) {
+      const data = parseStdoutJson<{ valid: boolean; errors: unknown[] }>(stdout);
+      expect(data.valid).toBe(false);
+      expect(Array.isArray(data.errors)).toBe(true);
+    }
+  });
+
+  test("dataset upload --no-validate --dry-run 跳过本地校验", async () => {
+    const file = join(__dirname, ".dataset-invalid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "dataset",
+      "upload",
+      "--file",
+      file,
+      "--no-validate",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ action: string; validate: boolean }>(stdout);
+    expect(data.action).toBe("dataset.upload");
+    expect(data.validate).toBe(false);
+  });
+
+  test("dataset validate 自动识别 DPO 并校验 chosen/rejected", async () => {
+    // No --schema: a record carrying chosen/rejected is auto-detected as DPO
+    // and the valid fixture passes.
+    const file = join(__dirname, ".dataset-dpo-valid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "dataset",
+      "validate",
+      "--file",
+      file,
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ valid: boolean; stats: { totalRecords?: number } }>(stdout);
+    expect(data.valid).toBe(true);
+    expect(data.stats.totalRecords).toBe(2);
+  });
+
+  test("dataset validate 自动识别 CPT 并校验 {text} 记录", async () => {
+    // No --schema: a record carrying `text` (and no `messages`) is auto-detected
+    // as CPT and the valid fixture passes.
+    const file = join(__dirname, ".dataset-cpt-valid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "dataset",
+      "validate",
+      "--file",
+      file,
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ valid: boolean; stats: { totalRecords?: number } }>(stdout);
+    expect(data.valid).toBe(true);
+    expect(data.stats.totalRecords).toBe(2);
+  });
+
+  test("dataset validate --schema cpt 拒绝缺失 text 的记录", async () => {
+    const file = join(__dirname, ".dataset-valid.jsonl"); // SFT {messages}, no text
+    const { stdout, exitCode } = await runCli([
+      "dataset",
+      "validate",
+      "--file",
+      file,
+      "--schema",
+      "cpt",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode).not.toBe(0);
+    const data = parseStdoutJson<{ valid: boolean; errors: { code: string; path?: string }[] }>(
+      stdout,
+    );
+    expect(data.valid).toBe(false);
+    expect(data.errors.map((e) => e.code)).toContain("MISSING_TEXT");
+  });
+
+  test("dataset validate --schema dpo 拒绝缺失 rejected 的记录", async () => {
+    const file = join(__dirname, ".dataset-dpo-invalid.jsonl");
+    const { stdout, exitCode } = await runCli([
+      "dataset",
+      "validate",
+      "--file",
+      file,
+      "--schema",
+      "dpo",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode).not.toBe(0);
+    const data = parseStdoutJson<{ valid: boolean; errors: { code: string; path?: string }[] }>(
+      stdout,
+    );
+    expect(data.valid).toBe(false);
+    expect(data.errors.map((e) => e.code)).toContain("MISSING_REJECTED");
+  });
+
+  test("dataset validate --schema chatml 忽略 chosen/rejected（不报 DPO 错误）", async () => {
+    // Same invalid-DPO file, but --schema chatml must not run DPO checks.
+    const file = join(__dirname, ".dataset-dpo-invalid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "dataset",
+      "validate",
+      "--file",
+      file,
+      "--schema",
+      "chatml",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ valid: boolean; errors: { code: string }[] }>(stdout);
+    expect(data.valid).toBe(true);
+    expect(data.errors.filter((c) => c.code.startsWith("MISSING_"))).toEqual([]);
+  });
+
+  test("dataset validate --schema <bad> 以非零码退出", async () => {
+    const file = join(__dirname, ".dataset-valid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "dataset",
+      "validate",
+      "--file",
+      file,
+      "--schema",
+      "sft",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode).not.toBe(0);
+    expect(`${stdout}\n${stderr}`).toMatch(/Unsupported --schema/);
+  });
+
+  test("dataset upload --dry-run 转发 --schema", async () => {
+    const file = join(__dirname, ".dataset-dpo-valid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "dataset",
+      "upload",
+      "--file",
+      file,
+      "--schema",
+      "dpo",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ action: string; schema: string }>(stdout);
+    expect(data.action).toBe("dataset.upload");
+    expect(data.schema).toBe("dpo");
+  });
+});
+
+describe.skipIf(!isDashScopeE2EReady())("e2e: dataset (DashScope)", () => {
+  test("dataset list --output json 返回结构化结果", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "dataset",
+      "list",
+      "--page-size",
+      "5",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ data?: { files?: unknown[] } }>(stdout);
+    expect(data).toBeTruthy();
+    if (data.data?.files) {
+      expect(Array.isArray(data.data.files)).toBe(true);
+    }
+  }, 60_000);
+});

packages/cli/tests/e2e/deploy.e2e.test.ts

@@ -0,0 +1,168 @@
+import { describe, expect, test } from "vite-plus/test";
+import { isDashScopeE2EReady, parseStdoutJson, runCli } from "./helpers.ts";
+
+/**
+ * Deploy E2E.
+ *
+ * The suite exercises command discovery, help text, and the `--dry-run`
+ * structured-output path (arg parsing + body construction) with no network
+ * dependency. Because `ensureApiKey` runs before every command (see main.ts),
+ * these cases are gated by isDashScopeE2EReady() — they are skipped when no
+ * DashScope credential is present (e.g. on CI) and run offline when one is.
+ * The remote list test is also gated and tolerates both empty accounts and
+ * auth/permission failures (see the test comment).
+ */
+
+describe.skipIf(!isDashScopeE2EReady())("e2e: deploy (offline)", () => {
+  test("deploy 列出子命令", async () => {
+    const { stdout, stderr, exitCode } = await runCli(["deploy"]);
+    expect(exitCode, stderr).toBe(0);
+    const out = `${stdout}\n${stderr}`;
+    expect(out).toMatch(/create|list|get|delete|update|scale|models/);
+  });
+
+  test("deploy create --help 正常退出并展示必填项", async () => {
+    const { stderr, exitCode } = await runCli(["deploy", "create", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--model|--name/i);
+  });
+
+  test("deploy create --dry-run 构造 lora 部署请求体", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "deploy",
+      "create",
+      "--model",
+      "qwen-plus-2025-12-01",
+      "--name",
+      "my-qwen-plus",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      action: string;
+      body: {
+        model_name: string;
+        name: string;
+        plan: string;
+        capacity: number;
+      };
+    }>(stdout);
+    expect(data.action).toBe("deploy.create");
+    expect(data.body.model_name).toBe("qwen-plus-2025-12-01");
+    expect(data.body.name).toBe("my-qwen-plus");
+    expect(data.body.plan).toBe("lora");
+    expect(data.body.capacity).toBe(1);
+  });
+
+  test("deploy scale --dry-run 转发 capacity", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "deploy",
+      "scale",
+      "--deployed-model",
+      "dep-xxx",
+      "--capacity",
+      "8",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      action: string;
+      deployed_model: string;
+      body: { capacity: number };
+    }>(stdout);
+    expect(data.action).toBe("deploy.scale");
+    expect(data.deployed_model).toBe("dep-xxx");
+    expect(data.body.capacity).toBe(8);
+  });
+
+  test("deploy update --dry-run 转发 rate limits", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "deploy",
+      "update",
+      "--deployed-model",
+      "dep-xxx",
+      "--rpm-limit",
+      "1000",
+      "--tpm-limit",
+      "200000",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      action: string;
+      body: { rpm_limit: number; tpm_limit: number };
+    }>(stdout);
+    expect(data.action).toBe("deploy.update");
+    expect(data.body.rpm_limit).toBe(1000);
+    expect(data.body.tpm_limit).toBe(200000);
+  });
+
+  test("deploy scale --dry-run 缺少 capacity/input-tpm/output-tpm 时报错", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "deploy",
+      "scale",
+      "--deployed-model",
+      "dep-xxx",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).not.toBe(0);
+    // Nothing useful emitted to stdout on a usage error.
+    expect(stdout.trim()).toBe("");
+  });
+
+  test.each([
+    ["list", ["--status", "RUNNING"]],
+    ["get", ["--deployed-model", "dep-xxx"]],
+    ["models", ["--source", "custom"]],
+    ["delete", ["--deployed-model", "dep-xxx"]],
+  ])("deploy %s --dry-run 发出结构化动作", async (sub, extra) => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "deploy",
+      sub,
+      ...extra,
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ action: string }>(stdout);
+    expect(data.action).toBe(`deploy.${sub}`);
+  });
+});
+
+describe.skipIf(!isDashScopeE2EReady())("e2e: deploy (DashScope)", () => {
+  /**
+   * 不同开发者的 key 状态不一：可能鉴权失败、可能账号下没有任何部署记录、
+   * 也可能受区域/权限限制。因此本用例不假设"有数据"或"调用成功"：
+   *   - 成功（exit 0）：响应必须可解析；deployments 可能为空数组或不存在。
+   *   - 失败（非零退出）：只要 CLI 把服务端/鉴权错误优雅上抛（stderr 有内容、
+   *     而非进程崩溃），即视为通过。
+   */
+  test("deploy list --output json 优雅返回（空账号或鉴权失败均通过）", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "deploy",
+      "list",
+      "--page-size",
+      "5",
+      "--output",
+      "json",
+    ]);
+    if (exitCode === 0) {
+      const data = parseStdoutJson<{ data?: { deployments?: unknown[] } }>(stdout);
+      expect(data).toBeTruthy();
+      if (data.data?.deployments) {
+        expect(Array.isArray(data.data.deployments)).toBe(true);
+      }
+    } else {
+      expect(stderr.length).toBeGreaterThan(0);
+    }
+  }, 60_000);
+});

packages/cli/tests/e2e/finetune.e2e.test.ts

@@ -0,0 +1,296 @@
+import { describe, expect, test } from "vite-plus/test";
+import { join } from "path";
+import { isDashScopeE2EReady, parseStdoutJson, runCli, cliPackageRoot } from "./helpers.ts";
+
+/**
+ * Fine-tune E2E.
+ *
+ * The suite exercises command discovery, help text, and the `--dry-run`
+ * structured-output path (arg parsing + body construction) with no network
+ * dependency. Because `ensureApiKey` runs before every command (see main.ts),
+ * these cases are gated by isDashScopeE2EReady() — they are skipped when no
+ * DashScope credential is present (e.g. on CI) and run offline when one is.
+ * The remote list test is also gated and tolerates both empty accounts and
+ * auth/permission failures (see the test comment).
+ */
+
+describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (offline)", () => {
+  test("finetune 列出子命令", async () => {
+    const { stdout, stderr, exitCode } = await runCli(["finetune"]);
+    expect(exitCode, stderr).toBe(0);
+    const out = `${stdout}\n${stderr}`;
+    expect(out).toMatch(/create|list|get|cancel|delete|logs|checkpoints|export|watch|capability/);
+  });
+
+  test("finetune create --help 正常退出并展示必填项", async () => {
+    const { stderr, exitCode } = await runCli(["finetune", "create", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--model|--datasets/i);
+  });
+
+  test("finetune create --dry-run 构造 SFT 默认请求体", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "create",
+      "--model",
+      "qwen3-8b",
+      "--datasets",
+      "file-aaa,file-bbb",
+      "--validations",
+      "file-ccc",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      action: string;
+      body: {
+        model: string;
+        training_file_ids: string[];
+        validation_file_ids: string[];
+        training_type: string;
+        hyper_parameters: { n_epochs: number };
+      };
+    }>(stdout);
+    expect(data.action).toBe("finetune.create");
+    expect(data.body.model).toBe("qwen3-8b");
+    expect(data.body.training_file_ids).toEqual(["file-aaa", "file-bbb"]);
+    expect(data.body.validation_file_ids).toEqual(["file-ccc"]);
+    expect(data.body.training_type).toBe("efficient_sft");
+    expect(data.body.hyper_parameters.n_epochs).toBe(3);
+  });
+
+  test("finetune create --dry-run 转发训练类型与超参", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "create",
+      "--model",
+      "qwen3-8b",
+      "--datasets",
+      "file-aaa",
+      "--training-type",
+      "sft-lora",
+      "--n-epochs",
+      "5",
+      "--batch-size",
+      "16",
+      "--learning-rate",
+      "1.6e-5",
+      "--max-length",
+      "4096",
+      "--model-name",
+      "my-qwen-sft",
+      "--suffix",
+      "v1",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      action: string;
+      body: {
+        training_type: string;
+        model_name: string;
+        finetuned_output_suffix: string;
+        hyper_parameters: {
+          n_epochs: number;
+          batch_size: number;
+          learning_rate: string;
+          max_length: number;
+        };
+      };
+    }>(stdout);
+    expect(data.body.training_type).toBe("efficient_sft");
+    expect(data.body.model_name).toBe("my-qwen-sft");
+    expect(data.body.finetuned_output_suffix).toBe("v1");
+    // batch_size is forwarded verbatim when within the [8, 1024] server range.
+    expect(data.body.hyper_parameters).toEqual({
+      n_epochs: 5,
+      batch_size: 16,
+      learning_rate: "1.6e-5",
+      max_length: 4096,
+    });
+  });
+
+  test("finetune create --training-type 拒绝不支持的训练类型值", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "create",
+      "--model",
+      "qwen3-8b",
+      "--datasets",
+      "file-aaa",
+      "--training-type",
+      "cpt-lora",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stdout + stderr).not.toBe(0);
+  });
+
+  test("finetune create --dry-run 把本地路径标记为 pending 上传且不发起网络请求", async () => {
+    const localPath = join(cliPackageRoot, "tests", "e2e", ".dataset-valid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "create",
+      "--model",
+      "qwen3-8b",
+      "--datasets",
+      `${localPath},file-bbb`,
+      "--validations",
+      localPath,
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      action: string;
+      body: { training_file_ids: string[]; validation_file_ids: string[] };
+      pending_uploads: { field: string; path: string }[];
+    }>(stdout);
+    expect(data.action).toBe("finetune.create");
+    // Local path preserved verbatim in the body (no upload in dry-run).
+    expect(data.body.training_file_ids[0]).toBe(localPath);
+    expect(data.body.training_file_ids[1]).toBe("file-bbb");
+    expect(data.body.validation_file_ids).toEqual([localPath]);
+    // Two pending uploads: training (1 local) + validation (1 local).
+    expect(data.pending_uploads).toHaveLength(2);
+    expect(data.pending_uploads.map((p) => p.field).sort()).toEqual(["datasets", "validations"]);
+  });
+
+  test("finetune create --datasets 为空时拒绝", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "create",
+      "--model",
+      "qwen3-8b",
+      "--datasets",
+      " , ",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stdout + stderr).not.toBe(0);
+  });
+
+  test("finetune create 样本数 <= batch_size 时提交前快速失败且不上传", async () => {
+    // The fixture has 3 records; the small-file auto-adjust sets batch_size=8,
+    // so 3 <= 8 trips the pre-submit gate. The gate fires before any upload,
+    // so this is fully offline (no key, no network) — the proof is that the
+    // error is the gate message AND no "Uploaded …" line ever appears.
+    const localPath = join(cliPackageRoot, "tests", "e2e", ".dataset-valid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "create",
+      "--model",
+      "qwen3-8b",
+      "--datasets",
+      localPath,
+      "--yes",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stdout + stderr).not.toBe(0);
+    const combined = `${stdout}\n${stderr}`;
+    expect(combined).toMatch(/not greater than batch_size/i);
+    // Crucially, no upload happened — the gate must fire before the upload step.
+    expect(combined).not.toMatch(/Uploaded .* → file-/);
+  });
+
+  test("finetune create --batch-size 过小仍按 8 下限比较（不绕过卡口）", async () => {
+    // Even with --batch-size 1 (server clamps to 8), 3 samples <= 8 still trips
+    // the gate — confirms the gate uses the clamped/effective batch, not the raw.
+    const localPath = join(cliPackageRoot, "tests", "e2e", ".dataset-valid.jsonl");
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "create",
+      "--model",
+      "qwen3-8b",
+      "--datasets",
+      localPath,
+      "--batch-size",
+      "1",
+      "--yes",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stdout + stderr).not.toBe(0);
+    expect(`${stdout}\n${stderr}`).toMatch(/batch_size \(8\)/);
+  });
+
+  test.each([
+    ["list", ["--status", "RUNNING"]],
+    ["get", ["--job-id", "ft-xxx"]],
+    ["checkpoints", ["--job-id", "ft-xxx"]],
+    ["logs", ["--job-id", "ft-xxx", "--page-size", "50"]],
+    ["export", ["--job-id", "ft-xxx", "--checkpoint", "ckpt-3", "--model-name", "m"]],
+    ["cancel", ["--job-id", "ft-xxx"]],
+    ["delete", ["--job-id", "ft-xxx"]],
+    ["watch", ["--job-id", "ft-xxx"]],
+    ["capability", ["--model", "qwen3-8b"]],
+  ])("finetune %s --dry-run 发出结构化动作", async (sub, extra) => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      sub,
+      ...extra,
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ action: string }>(stdout);
+    expect(data.action).toBe(`finetune.${sub}`);
+  });
+
+  test("finetune create --dry-run 解析多 datasets 中的空白", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "create",
+      "--model",
+      "qwen3-8b",
+      "--datasets",
+      " file-a , ,file-b ",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      body: { training_file_ids: string[] };
+    }>(stdout);
+    expect(data.body.training_file_ids).toEqual(["file-a", "file-b"]);
+  });
+});
+
+describe.skipIf(!isDashScopeE2EReady())("e2e: finetune (DashScope)", () => {
+  /**
+   * 不同开发者的 key 状态不一：可能鉴权失败、可能账号下没有任何微调记录、
+   * 也可能受区域/权限限制。因此本用例不假设"有数据"或"调用成功"：
+   *   - 成功（exit 0）：响应必须可解析；jobs 可能为空数组或不存在。
+   *   - 失败（非零退出）：只要 CLI 把服务端/鉴权错误优雅上抛（stderr 有内容、
+   *     而非进程崩溃），即视为通过。
+   */
+  test("finetune list --output json 优雅返回（空账号或鉴权失败均通过）", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "finetune",
+      "list",
+      "--page-size",
+      "5",
+      "--output",
+      "json",
+    ]);
+    if (exitCode === 0) {
+      const data = parseStdoutJson<{ data?: { jobs?: unknown[] } }>(stdout);
+      expect(data).toBeTruthy();
+      if (data.data?.jobs) {
+        expect(Array.isArray(data.data.jobs)).toBe(true);
+      }
+    } else {
+      expect(stderr.length).toBeGreaterThan(0);
+    }
+  }, 60_000);
+});

packages/cli/tests/e2e/helpers.ts

@@ -101,6 +101,26 @@ export function isDashScopeE2EReady(): boolean {
   }
 }
 
+/**
+ * Console-gateway 命令（quota / usage free / usage stats）的 E2E 就绪检查：
+ * 需 `BAILIAN_E2E=1` 且存在 console access_token（环境变量 `DASHSCOPE_ACCESS_TOKEN`
+ * 或 `~/.bailian/config.json` 的 `access_token`）。
+ *
+ * 仅检查 token 是否存在——无法本地判断是否过期。token 过期时 gated 用例仍会执行，
+ * 但用 `isConsoleAuthFailure` 把“session 未登录/已过期”的优雅报错视为通过，保持
+ * 与 deploy/dataset “无 key / 有效 key / 失效 key 均绿”的一致策略。
+ */
+export function isConsoleE2EReady(): boolean {
+  if (!isBailianE2EEnabled()) return false;
+  if (process.env.DASHSCOPE_ACCESS_TOKEN?.trim()) return true;
+  try {
+    const config = readConfigFile();
+    return typeof config.access_token === "string" && config.access_token.length > 0;
+  } catch {
+    return false;
+  }
+}
+
 /** 语音与图像（可设 `BAILIAN_E2E_MEDIA=0` 在仅跑文本/记忆/知识库时跳过） */
 export function isBailianE2EMediaEnabled(): boolean {
   if (process.env.BAILIAN_E2E_MEDIA === "0") return false;
@@ -117,8 +137,17 @@ export function e2eLabelFromMetaUrl(metaUrl: string): string {
   return basename(fileURLToPath(metaUrl), ".ts").replace(/\.e2e\.test$/, "");
 }
 
-/** 知识库用例：须显式索引 ID + AK/SK（workspace 可读 config / env，故不在此强制校验） */
+/** 知识库用例：须显式索引 ID + API-KEY 或 AK/SK */
 export function isKnowledgeE2EReady(): boolean {
+  if (!isBailianE2EEnabled()) return false;
+  if (!process.env.BAILIAN_E2E_INDEX_ID) return false;
+  const hasApiKey = isDashScopeE2EReady();
+  const hasAkSk =
+    !!process.env.ALIBABA_CLOUD_ACCESS_KEY_ID && !!process.env.ALIBABA_CLOUD_ACCESS_KEY_SECRET;
+  return hasApiKey || hasAkSk;
+}
+
+export function isKnowledgeAkSkReady(): boolean {
   return (
     isBailianE2EEnabled() &&
     !!process.env.ALIBABA_CLOUD_ACCESS_KEY_ID &&
@@ -172,3 +201,16 @@ export function parseStdoutJson<T = unknown>(stdout: string): T {
   const t = stdout.trim();
   return JSON.parse(t) as T;
 }
+
+/**
+ * 判断一次 CLI 运行是否因 console session 未登录/已过期而失败。
+ *
+ * Console E2E 用例的 readiness 闸（`isConsoleE2EReady`）只能判断 token 是否存在，
+ * 无法判断是否过期；token 失效时 gated 用例仍会执行并拿到鉴权错误。本函数让用例
+ * 参考 deploy/dataset 的做法：只要 CLI 把鉴权错误优雅上抛（非零退出 + stderr 说明
+ * session 失效），即视为通过，而不是强求 exit 0 的成功输出。
+ */
+export function isConsoleAuthFailure(result: RunCliResult): boolean {
+  if (result.exitCode === 0) return false;
+  return /not logged in|has expired|NotLogined|Run `bl auth login/i.test(result.stderr);
+}

packages/cli/tests/e2e/knowledge.e2e.test.ts

@@ -1,55 +1,206 @@
-import { join } from "path";
+import { tmpdir } from "os";
 import { describe, expect, test } from "vite-plus/test";
-import {
-  isBailianE2EEnabled,
-  isKnowledgeE2EReady,
-  monorepoRoot,
-  parseStdoutJson,
-  runCli,
-} from "./helpers.ts";
+import { parseStdoutJson, runCli } from "./helpers.ts";
 
-// 已开启 E2E 但 AK/SK、索引等未齐时提醒配置根目录 .env（否则本文件整组 describe 会被 skip）
-if (isBailianE2EEnabled() && !isKnowledgeE2EReady()) {
-  const envFile = join(monorepoRoot(), ".env");
-  console.warn(
-    [
-      "[e2e:knowledge] 知识库检索需要 RAM 的 AK/SK、索引 ID，以及工作空间 ID；当前未就绪，本组用例将被跳过。",
-      `请在 monorepo 根目录的 .env 中配置（${envFile}）：`,
-      "  ALIBABA_CLOUD_ACCESS_KEY_ID",
-      "  ALIBABA_CLOUD_ACCESS_KEY_SECRET",
-      "  BAILIAN_E2E_INDEX_ID",
-      "  BAILIAN_WORKSPACE_ID（也可执行: bl config set workspace_id <工作空间 id>）",
-    ].join("\n"),
-  );
+// ---- Types ----
+
+interface DryRunBody {
+  endpoint?: string;
+  request?: {
+    index_id?: string;
+    query?: string;
+    search_filters?: unknown[];
+    rerank_top_n?: number;
+    enable_reranking?: boolean;
+    dense_similarity_top_k?: number;
+    sparse_similarity_top_k?: number;
+    rerank?: Array<{ model_name?: string; rerank_mode?: string; rerank_instruct?: string }>;
+  };
 }
 
-interface KnowledgeRetrieveBody {
-  Success?: boolean;
-  Code?: string;
-  Data?: { Nodes?: unknown[] };
-}
+// ---- Help & missing args (no credentials needed) ----
 
-/** 知识库检索（需 AK/SK + workspace + 索引；未就绪则整组跳过） */
-describe.skipIf(!isKnowledgeE2EReady())("e2e: knowledge retrieve", () => {
-  test("知识库检索", async () => {
-    const indexId = process.env.BAILIAN_E2E_INDEX_ID!;
-    const { stdout, stderr, exitCode } = await runCli([
+describe("e2e: knowledge retrieve", () => {
+  test("knowledge 分组展示子命令帮助且成功退出", async () => {
+    const { stdout, stderr, exitCode } = await runCli(["knowledge"]);
+    expect(exitCode, stderr).toBe(0);
+    const out = `${stdout}\n${stderr}`;
+    expect(out).toMatch(/knowledge|retrieve/i);
+  });
+
+  test("knowledge retrieve --help 正常退出", async () => {
+    const { stderr, exitCode } = await runCli(["knowledge", "retrieve", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--index-id/i);
+    expect(stderr).toMatch(/--query/i);
+    expect(stderr).toMatch(/--rerank-top-n/i);
+    expect(stderr).toMatch(/deprecated/i);
+    expect(stderr).toMatch(/--workspace-id/i);
+  });
+
+  test("缺少 --index-id 时打印帮助并退出 (0)", async () => {
+    const { stderr, exitCode } = await runCli([
+      "knowledge",
+      "retrieve",
+      "--query",
+      "test",
+      "--non-interactive",
+    ]);
+    expect(exitCode).toBe(0);
+    expect(stderr).toMatch(/--index-id|Usage:/i);
+  });
+
+  test("缺少 --query 时打印帮助并退出 (0)", async () => {
+    const { stderr, exitCode } = await runCli([
       "knowledge",
       "retrieve",
       "--index-id",
-      indexId,
-      "--query",
-      "端到端检索测试",
-      "--top-k",
-      "3",
+      "idx_test",
       "--non-interactive",
-      "--output",
-      "json",
     ]);
-    expect(exitCode, stderr).toBe(0);
-    const data = parseStdoutJson<KnowledgeRetrieveBody>(stdout);
-    const ok = data.Success === true || data.Code === "Success";
-    expect(ok).toBe(true);
-    expect(Array.isArray(data.Data?.Nodes)).toBe(true);
-  }, 120_000);
+    expect(exitCode).toBe(0);
+    expect(stderr).toMatch(/--query|Usage:/i);
+  });
+});
+
+// ---- Error scenarios (no real credentials needed) ----
+
+describe("e2e: knowledge retrieve errors", () => {
+  test("无任何凭证时提示 No credentials found 并非零退出", async () => {
+    const { stderr, exitCode } = await runCli(
+      [
+        "knowledge",
+        "retrieve",
+        "--index-id",
+        "idx_test",
+        "--query",
+        "test",
+        "--non-interactive",
+        "--output",
+        "json",
+      ],
+      {
+        DASHSCOPE_API_KEY: undefined,
+        DASHSCOPE_ACCESS_TOKEN: undefined,
+        ALIBABA_CLOUD_ACCESS_KEY_ID: undefined,
+        ALIBABA_CLOUD_ACCESS_KEY_SECRET: undefined,
+        BAILIAN_CONFIG_DIR: tmpdir(),
+      },
+    );
+    expect(exitCode).not.toBe(0);
+    expect(stderr).toMatch(/no credentials found/i);
+  });
+});
+
+// ---- Dry-run (no real credentials needed) ----
+
+describe("e2e: knowledge retrieve dry-run", () => {
+  test("--dry-run 输出 endpoint 和 snake_case body", async () => {
+    const { stdout, stderr, exitCode } = await runCli(
+      [
+        "knowledge",
+        "retrieve",
+        "--dry-run",
+        "--index-id",
+        "idx_test",
+        "--query",
+        "hello",
+        "--non-interactive",
+        "--output",
+        "json",
+      ],
+      { DASHSCOPE_API_KEY: "sk-fake-for-dryrun" },
+    );
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<DryRunBody>(stdout);
+    expect(data.endpoint).toMatch(/api\/v1\/indices\/rag\/index\/retrieve/);
+    expect(data.request?.index_id).toBe("idx_test");
+    expect(data.request?.query).toBe("hello");
+  });
+
+  test("--dry-run + --top-k 转发到 rerank_top_n 并输出废弃警告", async () => {
+    const { stdout, stderr, exitCode } = await runCli(
+      [
+        "knowledge",
+        "retrieve",
+        "--dry-run",
+        "--index-id",
+        "idx_test",
+        "--query",
+        "hello",
+        "--top-k",
+        "5",
+        "--non-interactive",
+        "--output",
+        "json",
+      ],
+      { DASHSCOPE_API_KEY: "sk-fake-for-dryrun" },
+    );
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--top-k.*deprecated/i);
+    const data = parseStdoutJson<DryRunBody>(stdout);
+    expect(data.request?.rerank_top_n).toBe(5);
+  });
+
+  test("--dry-run + --rerank-top-n 优先于 --top-k", async () => {
+    const { stdout, stderr, exitCode } = await runCli(
+      [
+        "knowledge",
+        "retrieve",
+        "--dry-run",
+        "--index-id",
+        "idx_test",
+        "--query",
+        "hello",
+        "--top-k",
+        "5",
+        "--rerank-top-n",
+        "10",
+        "--non-interactive",
+        "--output",
+        "json",
+      ],
+      { DASHSCOPE_API_KEY: "sk-fake-for-dryrun" },
+    );
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<DryRunBody>(stdout);
+    expect(data.request?.rerank_top_n).toBe(10);
+  });
+
+  test("--dry-run + rerank 参数完整输出", async () => {
+    const { stdout, stderr, exitCode } = await runCli(
+      [
+        "knowledge",
+        "retrieve",
+        "--dry-run",
+        "--index-id",
+        "idx_test",
+        "--query",
+        "hello",
+        "--rerank",
+        "--rerank-model",
+        "qwen3-rerank-hybrid",
+        "--rerank-mode",
+        "custom",
+        "--rerank-instruct",
+        "按相关性排序",
+        "--dense-similarity-top-k",
+        "100",
+        "--sparse-similarity-top-k",
+        "50",
+        "--non-interactive",
+        "--output",
+        "json",
+      ],
+      { DASHSCOPE_API_KEY: "sk-fake-for-dryrun" },
+    );
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<DryRunBody>(stdout);
+    expect(data.request?.enable_reranking).toBe(true);
+    expect(data.request?.dense_similarity_top_k).toBe(100);
+    expect(data.request?.sparse_similarity_top_k).toBe(50);
+    expect(data.request?.rerank?.[0]?.model_name).toBe("qwen3-rerank-hybrid");
+    expect(data.request?.rerank?.[0]?.rerank_mode).toBe("custom");
+    expect(data.request?.rerank?.[0]?.rerank_instruct).toBe("按相关性排序");
+  });
 });

packages/cli/tests/e2e/mcp.e2e.test.ts

@@ -66,7 +66,7 @@ describe("e2e: mcp", () => {
     expect(exitCode, stderr).toBe(0);
     const data = parseStdoutJson<{
       api?: string;
-      region?: string;
+      consoleRegion?: string;
       data?: {
         reqDTO?: {
           type?: string;
@@ -79,7 +79,6 @@ describe("e2e: mcp", () => {
       };
     }>(stdout);
     expect(data.api).toBe("zeldaEasy.broadscope-bailian.mcp-server.PageList");
-    expect(data.region).toBe("cn-beijing");
     expect(data.data?.reqDTO?.activated).toBe(1);
     expect(data.data?.reqDTO?.displayTools).toBe(false);
     expect(data.data?.reqDTO?.type).toBe("OFFICIAL");
@@ -88,7 +87,7 @@ describe("e2e: mcp", () => {
     expect(data.data?.reqDTO?.pageSize).toBe(5);
   });
 
-  test("mcp list --dry-run 自定义 --region 透传", async () => {
+  test("mcp list --dry-run 自定义 --console-region 透传", async () => {
     const { stdout, stderr, exitCode } = await runCli([
       "mcp",
       "list",
@@ -96,12 +95,12 @@ describe("e2e: mcp", () => {
       "--non-interactive",
       "--output",
       "json",
-      "--region",
+      "--console-region",
       "cn-hangzhou",
     ]);
     expect(exitCode, stderr).toBe(0);
-    const data = parseStdoutJson<{ region?: string }>(stdout);
-    expect(data.region).toBe("cn-hangzhou");
+    const data = parseStdoutJson<{ consoleRegion?: string }>(stdout);
+    expect(data.consoleRegion).toBe("cn-hangzhou");
   });
 
   test("mcp tools <server-code> --dry-run 输出 /api/v1/mcps/<code>/mcp 形态 URL", async () => {

packages/cli/tests/e2e/omni.e2e.test.ts

@@ -0,0 +1,139 @@
+import { describe, expect, test } from "vite-plus/test";
+import { join } from "node:path";
+import {
+  e2eLabelFromMetaUrl,
+  isBailianE2EMediaEnabled,
+  isDashScopeE2EReady,
+  makeE2eOutputDir,
+  parseStdoutJson,
+  runCli,
+} from "./helpers.ts";
+
+describe("e2e: omni", () => {
+  test("omni --help 正常退出", async () => {
+    const { stderr, exitCode } = await runCli(["omni", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/omni|--message|--audio|text-only/i);
+  });
+});
+
+describe.skipIf(!isBailianE2EMediaEnabled() || !isDashScopeE2EReady())(
+  "e2e: omni（DashScope 媒体）",
+  () => {
+    test("omni --list-voices 输出音色列表并退出", async () => {
+      const { stdout, stderr, exitCode } = await runCli(["omni", "--list-voices"]);
+      expect(exitCode, stderr).toBe(0);
+      expect(stdout).toMatch(/Omni output voices:/);
+      expect(stdout).toMatch(/Tina/);
+      expect(stdout).toMatch(/Dylan/);
+      expect(stdout).toMatch(/Total: 13 voices/);
+    });
+    test("omni 缺少 --message 时打印子命令帮助并退出 (0)", async () => {
+      const { stderr, exitCode } = await runCli([
+        "omni",
+        "--model",
+        "qwen3.5-omni-flash",
+        "--non-interactive",
+      ]);
+      expect(exitCode).toBe(0);
+      expect(stderr).toMatch(/--message|Usage:/i);
+    });
+
+    test("omni --audio 无法识别扩展名时退出为用法错误 (2)", async () => {
+      const { stderr, exitCode } = await runCli([
+        "omni",
+        "--model",
+        "qwen3.5-omni-flash",
+        "--audio",
+        "https://example.com/sample.flac",
+        "--text-only",
+        "--message",
+        "这段音频在说什么？",
+        "--non-interactive",
+      ]);
+      expect(exitCode).toBe(2);
+      expect(stderr).toMatch(/Unsupported audio extension|Cannot infer audio format/i);
+    });
+
+    test("omni --dry-run --audio 构造 input_audio 而非 audio_url", async () => {
+      const { stdout, stderr, exitCode } = await runCli([
+        "omni",
+        "--dry-run",
+        "--model",
+        "qwen3.5-omni-flash",
+        "--audio",
+        "https://example.com/sample.wav",
+        "--text-only",
+        "--message",
+        "这段音频在说什么？",
+        "--non-interactive",
+        "--output",
+        "json",
+      ]);
+      expect(exitCode, stderr).toBe(0);
+      const data = parseStdoutJson<{
+        request?: {
+          messages?: Array<{
+            content?: Array<{
+              type?: string;
+              audio_url?: unknown;
+              input_audio?: { data?: string; format?: string };
+            }>;
+          }>;
+        };
+      }>(stdout);
+      const parts = data.request?.messages?.flatMap((m) =>
+        Array.isArray(m.content) ? m.content : [],
+      );
+      const audioPart = parts?.find((p) => p.type === "input_audio" || p.type === "audio_url");
+      expect(audioPart?.type).toBe("input_audio");
+      expect(audioPart?.audio_url).toBeUndefined();
+      expect(audioPart?.input_audio?.data).toBe("https://example.com/sample.wav");
+      expect(audioPart?.input_audio?.format).toBe("wav");
+    });
+
+    test("【qwen3.5-omni-flash】本地音频理解", async () => {
+      const outDir = makeE2eOutputDir(e2eLabelFromMetaUrl(import.meta.url));
+      const clipText = "端到端Omni音频测试";
+      const clipWav = join(outDir, "e2e-omni-input.wav");
+
+      const syn = await runCli([
+        "speech",
+        "synthesize",
+        "--model",
+        "cosyvoice-v3-flash",
+        "--voice",
+        "longxiaochun_v3",
+        "--text",
+        clipText,
+        "--format",
+        "wav",
+        "--out",
+        clipWav,
+        "--non-interactive",
+        "--output",
+        "json",
+      ]);
+      expect(syn.exitCode, syn.stderr).toBe(0);
+
+      const omni = await runCli([
+        "omni",
+        "--model",
+        "qwen3.5-omni-flash",
+        "--audio",
+        clipWav,
+        "--text-only",
+        "--system",
+        "请逐字转写用户提供的音频内容，不要添加解释。",
+        "--message",
+        "请转写这段音频。",
+        "--non-interactive",
+        "--output",
+        "json",
+      ]);
+      expect(omni.exitCode, omni.stderr).toBe(0);
+      const body = parseStdoutJson<{ content?: string }>(omni.stdout);
+      expect(body.content?.replace(/\s/g, "")).toMatch(/端到端Omni音频测试/);
+    }, 180_000);
+  },
+);

packages/cli/tests/e2e/proxy.e2e.test.ts

@@ -0,0 +1,124 @@
+import { execFile } from "child_process";
+import { createServer, type Server } from "http";
+import { mkdtempSync, rmSync, writeFileSync } from "fs";
+import type { AddressInfo } from "net";
+import { tmpdir } from "os";
+import { join } from "path";
+import { promisify } from "util";
+import { afterAll, beforeAll, describe, expect, test } from "vite-plus/test";
+import { cliPackageRoot } from "./helpers.ts";
+
+const execFileAsync = promisify(execFile);
+
+/**
+ * 代理支持 E2E（issue #35）：只验证 `setupProxyFromEnv()` 是否把代理 dispatcher
+ * 正确装到全局 fetch 上——设了 HTTPS_PROXY 后裸 `fetch()` 走代理，未设置时直连，
+ * NO_PROXY 命中时跳过，非法代理值给出明确报错。
+ *
+ * 不经过任何 CLI 命令（不解析凭证、不打 gateway），因此 CI 上无需 api key /
+ * access token，与既有 e2e 设计一致。全程离线：目标域名用 `.invalid`（保留顶级域，
+ * 必然无法解析），代理收到 CONNECT 后规范返回 502，不产生真实外网请求。
+ */
+
+const FAKE_HOST = "bl-proxy-e2e.invalid";
+const FAKE_URL = `https://${FAKE_HOST}/probe`;
+
+/**
+ * 最小探针脚本：调用真实的 `setupProxyFromEnv()`，再对目标发一个普通 fetch。
+ * 代理行为由进程环境变量决定，正是被测对象；fetch 成败不重要，我们只看代理是否收到 CONNECT。
+ */
+const PROBE_SCRIPT = `
+import { setupProxyFromEnv } from ${JSON.stringify(join(cliPackageRoot, "..", "runtime", "src", "proxy.ts"))};
+setupProxyFromEnv();
+try {
+  await fetch(${JSON.stringify(FAKE_URL)}, { signal: AbortSignal.timeout(5000) });
+} catch {
+  // 目标不可达/隧道被拒都正常——本测试只关心代理是否收到 CONNECT
+}
+`;
+
+let proxy: Server;
+let proxyUrl: string;
+let scriptDir: string;
+let scriptPath: string;
+const connectTargets: string[] = [];
+
+beforeAll(async () => {
+  proxy = createServer();
+  // 记录收到的 CONNECT 目标（host:port），并以 502 拒绝隧道
+  proxy.on("connect", (req, clientSocket) => {
+    connectTargets.push(req.url ?? "");
+    clientSocket.end("HTTP/1.1 502 Bad Gateway\r\n\r\n");
+  });
+  await new Promise<void>((resolve) => proxy.listen(0, "127.0.0.1", resolve));
+  proxyUrl = `http://127.0.0.1:${(proxy.address() as AddressInfo).port}`;
+
+  scriptDir = mkdtempSync(join(tmpdir(), "bl-proxy-e2e-"));
+  scriptPath = join(scriptDir, "probe.ts");
+  writeFileSync(scriptPath, PROBE_SCRIPT);
+});
+
+afterAll(async () => {
+  await new Promise<void>((resolve) => proxy.close(() => resolve()));
+  rmSync(scriptDir, { recursive: true, force: true });
+});
+
+/** 清空所有代理相关环境变量，确保每个用例只受自身设置影响 */
+const PROXY_ENV_CLEARED = {
+  HTTPS_PROXY: "",
+  https_proxy: "",
+  HTTP_PROXY: "",
+  http_proxy: "",
+  NO_PROXY: "",
+  no_proxy: "",
+};
+
+/** 以给定代理环境变量运行探针脚本，返回 { exitCode, stderr } */
+async function runProbe(
+  envOverrides: NodeJS.ProcessEnv,
+): Promise<{ exitCode: number; stderr: string }> {
+  try {
+    await execFileAsync("node", [scriptPath], {
+      cwd: cliPackageRoot,
+      encoding: "utf8",
+      env: { ...process.env, NODE_NO_WARNINGS: "1", ...PROXY_ENV_CLEARED, ...envOverrides },
+    });
+    return { exitCode: 0, stderr: "" };
+  } catch (err: unknown) {
+    const e = err as { stderr?: string; code?: number };
+    return { exitCode: typeof e.code === "number" ? e.code : 1, stderr: e.stderr ?? "" };
+  }
+}
+
+describe("e2e: proxy", () => {
+  test("设置 HTTPS_PROXY 后 fetch 经过代理（CONNECT 到目标主机）", async () => {
+    connectTargets.length = 0;
+    await runProbe({ HTTPS_PROXY: proxyUrl });
+    expect(connectTargets).toContain(`${FAKE_HOST}:443`);
+  });
+
+  test("空字符串小写变量不屏蔽大写 HTTPS_PROXY（undici ?? 取值回归）", async () => {
+    connectTargets.length = 0;
+    await runProbe({ https_proxy: "", HTTPS_PROXY: proxyUrl });
+    expect(connectTargets).toContain(`${FAKE_HOST}:443`);
+  });
+
+  test("NO_PROXY 命中目标主机时不走代理", async () => {
+    connectTargets.length = 0;
+    await runProbe({ HTTPS_PROXY: proxyUrl, NO_PROXY: FAKE_HOST });
+    expect(connectTargets.filter((t) => t.startsWith(FAKE_HOST))).toEqual([]);
+  });
+
+  test("未设置代理变量时保持直连（代理收不到任何流量）", async () => {
+    connectTargets.length = 0;
+    await runProbe({});
+    expect(connectTargets).toEqual([]);
+  });
+
+  test("代理 URL 非法时给出明确报错而非堆栈", async () => {
+    const { exitCode, stderr } = await runProbe({ HTTPS_PROXY: "::::not-a-url" });
+    expect(exitCode).not.toBe(0);
+    expect(stderr).toMatch(/Invalid proxy configuration/);
+    expect(stderr).toMatch(/HTTPS_PROXY/);
+  });
+});

packages/cli/tests/e2e/quota.e2e.test.ts

@@ -0,0 +1,282 @@
+import { describe, expect, test } from "vite-plus/test";
+import { isConsoleE2EReady, isConsoleAuthFailure, parseStdoutJson, runCli } from "./helpers.ts";
+
+describe("e2e: quota", () => {
+  test("quota list --help 正常退出", async () => {
+    const { stderr, exitCode } = await runCli(["quota", "list", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("--model");
+    expect(stderr).toContain("--all");
+  });
+
+  test("quota list --help 包含所有示例", async () => {
+    const { stderr, exitCode } = await runCli(["quota", "list", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("bl quota list");
+    expect(stderr).toContain("bl quota list --model qwen3.6-plus");
+    expect(stderr).toContain("bl quota list --all");
+  });
+
+  test("quota request --help 正常退出", async () => {
+    const { stderr, exitCode } = await runCli(["quota", "request", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("--model");
+    expect(stderr).toContain("--tpm");
+    expect(stderr).toContain("--yes");
+  });
+
+  test("quota history --help 正常退出", async () => {
+    const { stderr, exitCode } = await runCli(["quota", "history", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("--page");
+    expect(stderr).toContain("--model");
+  });
+
+  test("quota check --help 正常退出", async () => {
+    const { stderr, exitCode } = await runCli(["quota", "check", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("--model");
+    expect(stderr).toContain("--period");
+    expect(stderr).toContain("bl quota check");
+  });
+
+  test("quota check --period 0 报错最小值", async () => {
+    const { stderr, exitCode } = await runCli(["quota", "check", "--period", "0.5"]);
+    expect(exitCode).toBe(1);
+    expect(stderr).toContain("at least 1 minute");
+  });
+});
+
+describe.skipIf(!isConsoleE2EReady())("e2e: quota（Console）", () => {
+  test("quota list --dry-run 输出请求参数", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "quota",
+      "list",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      api?: string;
+      data?: {
+        input?: { queryQpmInfo?: boolean; supports?: { selfServiceLimitIncrease?: boolean } };
+      };
+    }>(stdout);
+    expect(data.api).toContain("listFoundationModels");
+    expect(data.data?.input?.queryQpmInfo).toBe(true);
+    expect(data.data?.input?.supports?.selfServiceLimitIncrease).toBe(true);
+  });
+
+  test("quota list --dry-run --all 不传 supports 过滤", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "quota",
+      "list",
+      "--all",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      data?: { input?: { supports?: unknown } };
+    }>(stdout);
+    expect(data.data?.input?.supports).toBeUndefined();
+  });
+
+  test("quota list 文本输出包含英文表头", async () => {
+    const result = await runCli(["quota", "list", "--output", "text", "--no-color"]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("quota list --model 指定模型返回结果", async () => {
+    const result = await runCli([
+      "quota",
+      "list",
+      "--model",
+      "qwen3.6-plus",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("quota list --model 不存在的模型报错", async () => {
+    const result = await runCli([
+      "quota",
+      "list",
+      "--model",
+      "nonexistent-model-xyz-99999",
+      "--output",
+      "text",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode).toBe(1);
+    expect(result.stderr).toContain("no matching models found");
+  });
+
+  test("quota list JSON 输出包含 model/rpm/tpm/maxTPM", async () => {
+    const result = await runCli(["quota", "list", "--output", "json"]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("quota request --dry-run 输出请求参数", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "quota",
+      "request",
+      "--model",
+      "qwen3.6-plus",
+      "--tpm",
+      "6000000",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      api?: string;
+      data?: { input?: { model?: string; limit?: { usage_limit?: number } } };
+    }>(stdout);
+    expect(data.api).toContain("updateFoundationModelLimits");
+    expect(data.data?.input?.model).toBe("qwen3.6-plus");
+    expect(data.data?.input?.limit?.usage_limit).toBeTypeOf("number");
+  });
+
+  test("quota request TPM 超范围报错", async () => {
+    const result = await runCli(["quota", "request", "--model", "qwen3.6-plus", "--tpm", "999"]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode).toBe(1);
+    expect(result.stderr).toContain("out of range");
+    expect(result.stderr).toContain("Current");
+    expect(result.stderr).toContain("Range");
+  });
+
+  test("quota request 不支持提额的模型报错", async () => {
+    const result = await runCli([
+      "quota",
+      "request",
+      "--model",
+      "nonexistent-model-xyz-99999",
+      "--tpm",
+      "100000",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode).toBe(1);
+    expect(result.stderr).toContain("not found");
+  });
+
+  test("quota history --dry-run 输出请求参数", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "quota",
+      "history",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      api?: string;
+      data?: { input?: { pageNo?: number; pageSize?: number } };
+    }>(stdout);
+    expect(data.api).toContain("listModelLimitApplications");
+    expect(data.data?.input?.pageNo).toBe(1);
+    expect(data.data?.input?.pageSize).toBe(10);
+  });
+
+  test("quota check --dry-run 输出 API 信息", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "quota",
+      "check",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ apis?: string[]; consoleRegion?: string }>(stdout);
+    expect(data.apis).toContain(
+      "zeldaHttp.dashscopeModel./zelda/api/v1/modelCenter/listFoundationModels",
+    );
+    expect(data.apis).toContain("zeldaEasy.bailian-telemetry.monitor.getMonitorData");
+    expect(data.consoleRegion).toBe("cn-beijing");
+  });
+
+  test("quota check --dry-run --console-region 透传", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "quota",
+      "check",
+      "--dry-run",
+      "--non-interactive",
+      "--output",
+      "json",
+      "--console-region",
+      "cn-hangzhou",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{ consoleRegion?: string }>(stdout);
+    expect(data.consoleRegion).toBe("cn-hangzhou");
+  });
+
+  test("quota check 文本输出包含英文表头", async () => {
+    const result = await runCli(["quota", "check", "--output", "text", "--no-color"]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("quota check --model 指定单模型", async () => {
+    const result = await runCli([
+      "quota",
+      "check",
+      "--model",
+      "qwen3.6-plus",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("quota check --model 逗号分隔多模型", async () => {
+    const result = await runCli([
+      "quota",
+      "check",
+      "--model",
+      "qwen3.6-plus,qwen-plus",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("quota check JSON 输出包含用量和限额字段", async () => {
+    const result = await runCli(["quota", "check", "--model", "qwen3.6-plus", "--output", "json"]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("quota history --dry-run --page 2 --page-size 20", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "quota",
+      "history",
+      "--page",
+      "2",
+      "--page-size",
+      "20",
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      data?: { input?: { pageNo?: number; pageSize?: number } };
+    }>(stdout);
+    expect(data.data?.input?.pageNo).toBe(2);
+    expect(data.data?.input?.pageSize).toBe(20);
+  });
+});

packages/cli/tests/e2e/usage-free.e2e.test.ts

@@ -0,0 +1,235 @@
+import { describe, expect, test } from "vite-plus/test";
+import { isConsoleE2EReady, isConsoleAuthFailure, parseStdoutJson, runCli } from "./helpers.ts";
+
+describe("e2e: usage free", () => {
+  test("usage 分组展示子命令帮助且退出码为 0", async () => {
+    const { stdout, stderr, exitCode } = await runCli(["usage"]);
+    expect(exitCode, stderr).toBe(0);
+    const out = `${stdout}\n${stderr}`;
+    expect(out).toMatch(/usage|free|freetier/i);
+  });
+
+  test("usage free --help 正常退出", async () => {
+    const { stderr, exitCode } = await runCli(["usage", "free", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--model|quota|free-tier/i);
+  });
+
+  test("usage free --help 包含所有示例", async () => {
+    const { stderr, exitCode } = await runCli(["usage", "free", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("bl usage free");
+    expect(stderr).toContain("bl usage free --model qwen3-max");
+    expect(stderr).toContain("bl usage free --model qwen3-max,qwen-turbo");
+  });
+});
+
+describe.skipIf(!isConsoleE2EReady())("e2e: usage free（Console）", () => {
+  test("usage free --dry-run --model 输出请求参数不发起调用", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "usage",
+      "free",
+      "--dry-run",
+      "--model",
+      "qwen3-max",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      api?: string;
+      data?: { queryFreeTierQuotaRequest?: { models?: string[] } };
+    }>(stdout);
+    expect(data.api).toContain("queryFreeTierQuota");
+    expect(data.data?.queryFreeTierQuotaRequest?.models).toEqual(["qwen3-max"]);
+  });
+
+  test("usage free --dry-run --model 逗号分隔多个模型", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "usage",
+      "free",
+      "--dry-run",
+      "--model",
+      "qwen3-max,qwen-turbo",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      data?: { queryFreeTierQuotaRequest?: { models?: string[] } };
+    }>(stdout);
+    expect(data.data?.queryFreeTierQuotaRequest?.models).toEqual(["qwen3-max", "qwen-turbo"]);
+  });
+
+  test("usage free --dry-run --model 重复模型名自动去重", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "usage",
+      "free",
+      "--dry-run",
+      "--model",
+      "qwen3-max,qwen3-max,qwen-turbo",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      data?: { queryFreeTierQuotaRequest?: { models?: string[] } };
+    }>(stdout);
+    expect(data.data?.queryFreeTierQuotaRequest?.models).toEqual(["qwen3-max", "qwen-turbo"]);
+  });
+
+  test("usage free --dry-run --model 逗号间有空格也能正确解析", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "usage",
+      "free",
+      "--dry-run",
+      "--model",
+      "qwen3-max, qwen-turbo",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      data?: { queryFreeTierQuotaRequest?: { models?: string[] } };
+    }>(stdout);
+    expect(data.data?.queryFreeTierQuotaRequest?.models).toEqual(["qwen3-max", "qwen-turbo"]);
+  });
+
+  test("usage free --dry-run 不指定 --model 传全量模型列表", async () => {
+    const { stderr, exitCode } = await runCli(["usage", "free", "--dry-run", "--output", "json"]);
+    expect(exitCode, stderr).toBe(0);
+  });
+
+  test("usage free --model 单模型查询返回 JSON 结果", async () => {
+    const result = await runCli(["usage", "free", "--model", "qwen3-max", "--output", "json"]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model 单模型文本输出包含表头", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "qwen3-max",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model 文本输出包含模型名", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "qwen3-max",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model 逗号分隔多模型文本输出包含所有模型", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "qwen3-max,qwen-turbo",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model 文本输出包含正确的 Type 列", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "qwen3-max",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model quotaStatus 为 UNKNOWN 时 Auto-Stop 显示 Unsupported", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "wan2.7-image",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model quotaStatus 为 UNKNOWN 时额度显示为 -", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "wan2.7-image",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model 不存在的模型仍返回表格行", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "nonexistent-model-xyz-12345",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model Auto-Stop 显示 ON、OFF 或 Unsupported", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "qwen3-max",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage free --model --console-region cn-beijing 指定区域查询", async () => {
+    const result = await runCli([
+      "usage",
+      "free",
+      "--model",
+      "qwen3-max",
+      "--console-region",
+      "cn-beijing",
+      "--output",
+      "json",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+});

packages/cli/tests/e2e/usage-stats.e2e.test.ts

@@ -0,0 +1,273 @@
+import { describe, expect, test } from "vite-plus/test";
+import { isConsoleE2EReady, isConsoleAuthFailure, parseStdoutJson, runCli } from "./helpers.ts";
+import { readConfigFile } from "bailian-cli-core";
+
+function getStaticWorkspaceId(): string | undefined {
+  if (process.env.BAILIAN_WORKSPACE_ID?.trim()) return process.env.BAILIAN_WORKSPACE_ID.trim();
+  try {
+    const config = readConfigFile();
+    if (config.workspace_id) return config.workspace_id;
+  } catch {}
+  return undefined;
+}
+
+// 当无静态 workspace-id 且 console 未登录/已过期时返回占位符，避免下游 dry-run
+// 用例因 `--workspace-id undefined` 而崩溃；live 用例各自用 isConsoleAuthFailure
+// 容忍鉴权失败。参考 deploy/dataset “无 key / 有效 / 失效 均绿”的策略。
+const FALLBACK_WORKSPACE_ID = "ws-e2e-unavailable";
+
+async function fetchDefaultWorkspaceId(): Promise<string> {
+  const staticId = getStaticWorkspaceId();
+  if (staticId) return staticId;
+
+  const result = await runCli(["workspace", "list", "--output", "json"]);
+  if (isConsoleAuthFailure(result) || result.exitCode !== 0) return FALLBACK_WORKSPACE_ID;
+  try {
+    const parsed = JSON.parse(result.stdout);
+    const data = parsed?.data?.DataV2?.data?.data?.data ?? [];
+    const defaultWs = data.find((ws: { defaultAgent?: boolean }) => ws.defaultAgent);
+    if (defaultWs?.workspaceId) return defaultWs.workspaceId;
+    if (data.length > 0 && data[0].workspaceId) return data[0].workspaceId;
+  } catch {
+    /* fall through to placeholder */
+  }
+  return FALLBACK_WORKSPACE_ID;
+}
+
+describe("e2e: usage stats", () => {
+  test("usage stats --help 正常退出", async () => {
+    const { stderr, exitCode } = await runCli(["usage", "stats", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toMatch(/--model|--days|stats/i);
+  });
+
+  test("usage stats --help 包含所有示例", async () => {
+    const { stderr, exitCode } = await runCli(["usage", "stats", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("bl usage stats");
+    expect(stderr).toContain("bl usage stats --model qwen-turbo");
+    expect(stderr).toContain("bl usage stats --days 30");
+  });
+
+  test("usage stats --help 包含 --workspace-id 选项", async () => {
+    const { stderr, exitCode } = await runCli(["usage", "stats", "--help"]);
+    expect(exitCode, stderr).toBe(0);
+    expect(stderr).toContain("--workspace-id");
+  });
+});
+
+describe.skipIf(!isConsoleE2EReady())("e2e: usage stats（Console）", () => {
+  let wsId: string;
+
+  test("获取默认 workspace-id", async () => {
+    wsId = await fetchDefaultWorkspaceId();
+    expect(wsId).toBeTypeOf("string");
+    expect(wsId.length).toBeGreaterThan(0);
+  });
+
+  test("usage stats --dry-run 概览模式输出请求参数", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--dry-run",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      api?: string;
+      data?: {
+        reqDTO?: {
+          startTime?: number;
+          endTime?: number;
+          modelCallSource?: string;
+          filterWorkspaceId?: string;
+        };
+      };
+    }>(stdout);
+    expect(data.api).toContain("getModelUsageStatistic");
+    expect(data.data?.reqDTO?.modelCallSource).toBe("Online");
+    expect(data.data?.reqDTO?.startTime).toBeTypeOf("number");
+    expect(data.data?.reqDTO?.endTime).toBeTypeOf("number");
+    expect(data.data?.reqDTO?.filterWorkspaceId).toBe(wsId);
+  });
+
+  test("usage stats --dry-run --days 30 时间跨度约 30 天", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--dry-run",
+      "--days",
+      "30",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      data?: { reqDTO?: { startTime?: number; endTime?: number } };
+    }>(stdout);
+    const span = (data.data?.reqDTO?.endTime ?? 0) - (data.data?.reqDTO?.startTime ?? 0);
+    const thirtyDaysMs = 30 * 24 * 60 * 60 * 1000;
+    expect(span).toBeGreaterThan(thirtyDaysMs - 5000);
+    expect(span).toBeLessThan(thirtyDaysMs + 5000);
+  });
+
+  test("usage stats --dry-run --model 指定模型使用 list API", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--dry-run",
+      "--model",
+      "qwen-turbo",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      api?: string;
+      data?: { reqDTO?: { model?: string; filterWorkspaceId?: string } };
+    }>(stdout);
+    expect(data.api).toContain("listModelUsageStatisticData");
+    expect(data.data?.reqDTO?.model).toBe("qwen-turbo");
+    expect(data.data?.reqDTO?.filterWorkspaceId).toBe(wsId);
+  });
+
+  test("usage stats --dry-run --type Text 传递 obsModelType", async () => {
+    const { stdout, stderr, exitCode } = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--dry-run",
+      "--type",
+      "Text",
+      "--output",
+      "json",
+    ]);
+    expect(exitCode, stderr).toBe(0);
+    const data = parseStdoutJson<{
+      data?: { reqDTO?: { obsModelType?: string } };
+    }>(stdout);
+    expect(data.data?.reqDTO?.obsModelType).toBe("Text");
+  });
+
+  test("usage stats 概览模式返回 JSON 结果", async () => {
+    const result = await runCli(["usage", "stats", "--workspace-id", wsId, "--output", "json"]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage stats 概览文本输出包含英文标签", async () => {
+    const result = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage stats 概览文本输出包含 Token 用量", async () => {
+    const result = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage stats --model 单模型文本输出包含英文表头", async () => {
+    const result = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--model",
+      "qwen3.6-plus",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage stats --model 逗号分隔多模型返回多行", async () => {
+    const result = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--model",
+      "qwen3.6-plus,deepseek-v4-pro",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage stats --model 不存在的模型返回空表格", async () => {
+    const result = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--model",
+      "nonexistent-model-xyz-99999",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage stats --days 1 短时间范围正常返回", async () => {
+    const result = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--days",
+      "1",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+
+  test("usage stats --type Vision 按类型过滤", async () => {
+    const result = await runCli([
+      "usage",
+      "stats",
+      "--workspace-id",
+      wsId,
+      "--type",
+      "Vision",
+      "--output",
+      "text",
+      "--no-color",
+    ]);
+    if (isConsoleAuthFailure(result)) return;
+    expect(result.exitCode, result.stderr).toBe(0);
+  });
+});

packages/cli/tests/e2e/video-download.e2e.test.ts

@@ -91,7 +91,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "video",
         "generate",
         "--model",
-        "happyhorse-1.0-t2v",
+        "happyhorse-1.1-t2v",
         "--duration",
         "3",
         "--prompt",

packages/cli/tests/e2e/video-generate-i2v.e2e.test.ts

@@ -37,7 +37,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "video",
         "generate",
         "--model",
-        "happyhorse-1.0-i2v",
+        "happyhorse-1.1-i2v",
         "--image",
         "https://example.com/placeholder.png",
         "--non-interactive",
@@ -53,7 +53,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "generate",
         "--dry-run",
         "--model",
-        "happyhorse-1.0-t2v",
+        "happyhorse-1.1-t2v",
         "--prompt",
         "干跑无图",
         "--non-interactive",
@@ -68,7 +68,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
       expect(data.request?.input?.media).toBeUndefined();
     });
 
-    test("【happyhorse-1.0-i2v】图片生成视频", async () => {
+    test("【happyhorse-1.1-i2v】图片生成视频", async () => {
       const outDir = makeE2eOutputDir(e2eLabelFromMetaUrl(import.meta.url));
       const png = join(outDir, "e2e-gen.png");
       const gen = await runCli([
@@ -95,7 +95,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "video",
         "generate",
         "--model",
-        "happyhorse-1.0-i2v",
+        "happyhorse-1.1-i2v",
         "--image",
         imagePath,
         "--prompt",

packages/cli/tests/e2e/video-generate-t2v.e2e.test.ts

@@ -37,7 +37,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "video",
         "generate",
         "--model",
-        "happyhorse-1.0-t2v",
+        "happyhorse-1.1-t2v",
         "--non-interactive",
       ]);
       expect(exitCode).toBe(0);
@@ -51,7 +51,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "generate",
         "--dry-run",
         "--model",
-        "happyhorse-1.0-t2v",
+        "happyhorse-1.1-t2v",
         "--prompt",
         "干跑校验",
         "--non-interactive",
@@ -62,18 +62,18 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
       const data = parseStdoutJson<{ request?: { model?: string; input?: { prompt?: string } } }>(
         stdout,
       );
-      expect(data.request?.model).toBe("happyhorse-1.0-t2v");
+      expect(data.request?.model).toBe("happyhorse-1.1-t2v");
       expect(data.request?.input?.prompt).toBe("干跑校验");
     });
 
-    test("【happyhorse-1.0-t2v】文本生成视频", async () => {
+    test("【happyhorse-1.1-t2v】文本生成视频", async () => {
       const outDir = makeE2eOutputDir(e2eLabelFromMetaUrl(import.meta.url));
       const { stdout, stderr, exitCode } = await runCli([
         ...cliTimeoutPrefix(),
         "video",
         "generate",
         "--model",
-        "happyhorse-1.0-t2v",
+        "happyhorse-1.1-t2v",
         "--prompt",
         "夕阳下海面波光，远景静态镜头",
         "--download",

packages/cli/tests/e2e/video-ref-r2v.e2e.test.ts

@@ -37,7 +37,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "video",
         "ref",
         "--model",
-        "happyhorse-1.0-r2v",
+        "happyhorse-1.1-r2v",
         "--image",
         "https://example.com/x.png",
         "--non-interactive",
@@ -52,7 +52,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "video",
         "ref",
         "--model",
-        "happyhorse-1.0-r2v",
+        "happyhorse-1.1-r2v",
         "--prompt",
         "仅有描述无素材",
         "--non-interactive",
@@ -61,7 +61,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
       expect(stderr).toMatch(/--image|ref-video|At least one|required/i);
     });
 
-    test("【happyhorse-1.0-r2v】视频参考生成", async () => {
+    test("【happyhorse-1.1-r2v】视频参考生成", async () => {
       const outDir = makeE2eOutputDir(e2eLabelFromMetaUrl(import.meta.url));
       const gen = await runCli([
         "image",
@@ -69,7 +69,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "--model",
         "qwen-image-2.0",
         "--prompt",
-        "一只简笔画小猫，白底",
+        "一片绿色的树叶，白底",
         "--out-dir",
         outDir,
         "--out-prefix",
@@ -88,7 +88,7 @@ describe.skipIf(!isBailianE2EVideoEnabled() || !isDashScopeE2EReady())(
         "video",
         "ref",
         "--model",
-        "happyhorse-1.0-r2v",
+        "happyhorse-1.1-r2v",
         "--prompt",
         "图1在画面中心轻微晃动",
         "--image",

packages/cli/tests/stress/lib/fixtures.mjs

@@ -180,7 +180,7 @@ export async function ensurePrerequisites(ctx) {
       "video",
       "generate",
       "--model",
-      "happyhorse-1.0-t2v",
+      "happyhorse-1.1-t2v",
       "--prompt",
       "压测前置短视频：海浪与静态远景，无明显人物。",
       "--duration",

packages/cli/tests/stress/lib/suite-fixtures.mjs

@@ -132,7 +132,7 @@ export async function generateCombinedFixtures({ suiteRoot, cliPackage }) {
       "video",
       "generate",
       "--model",
-      "happyhorse-1.0-t2v",
+      "happyhorse-1.1-t2v",
       "--prompt",
       "压测前置短视频：海浪与静态远景，无明显人物。",
       "--duration",

packages/cli/tests/stress/targets/video-i2v.mjs

@@ -16,7 +16,7 @@ const motions = [
 
 export const runStress = defineStressTarget({
   canonical: "video-i2v",
-  defaultModel: "happyhorse-1.0-i2v",
+  defaultModel: "happyhorse-1.1-i2v",
   batchDirPrefix: "video-i2v-batch",
   helpText: "pnpm run test:stress -- video-i2v [--reuse-fixtures] -- --count 5 -c 2",
 

packages/cli/tests/stress/targets/video-ref.mjs

@@ -16,7 +16,7 @@ const prompts = [
 
 export const runStress = defineStressTarget({
   canonical: "video-ref",
-  defaultModel: "happyhorse-1.0-r2v",
+  defaultModel: "happyhorse-1.1-r2v",
   batchDirPrefix: "video-ref-batch",
   helpText: "pnpm run test:stress -- video-ref [--reuse-fixtures] -- --count 5 -c 2",
 

packages/cli/tests/stress/targets/video-t2v.mjs

@@ -45,7 +45,7 @@ const pick = (arr) => arr[Math.floor(Math.random() * arr.length)];
 
 export const runStress = defineStressTarget({
   canonical: "video-t2v",
-  defaultModel: "happyhorse-1.0-t2v",
+  defaultModel: "happyhorse-1.1-t2v",
   batchDirPrefix: "video-t2v-batch",
   helpText: `用法：pnpm run test:stress -- video-t2v -- --concurrency 1 --count 3
 详见 docs/agents/stress-batch-tests.md`,

packages/cli/tsconfig.json

@@ -5,6 +5,7 @@
     "moduleDetection": "force",
     "module": "nodenext",
     "moduleResolution": "nodenext",
+    "customConditions": ["@bailian-cli/source"],
     "resolveJsonModule": true,
     "types": ["node"],
     "strict": true,

packages/commands/.gitignore

@@ -0,0 +1,4 @@
+node_modules
+dist
+*.log
+.DS_Store

packages/commands/package.json

@@ -0,0 +1,58 @@
+{
+  "name": "bailian-cli-commands",
+  "version": "1.5.0",
+  "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": {
+    "url": "https://github.com/modelstudioai/cli/issues"
+  },
+  "license": "Apache-2.0",
+  "author": "Aliyun Model Studio",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/modelstudioai/cli.git",
+    "directory": "packages/commands"
+  },
+  "files": [
+    "dist"
+  ],
+  "type": "module",
+  "types": "./dist/index.d.mts",
+  "exports": {
+    ".": {
+      "@bailian-cli/source": "./src/index.ts",
+      "default": "./dist/index.mjs"
+    },
+    "./package.json": "./package.json"
+  },
+  "publishConfig": {
+    "access": "public",
+    "exports": {
+      ".": "./dist/index.mjs",
+      "./package.json": "./package.json"
+    },
+    "registry": "https://registry.npmjs.org/"
+  },
+  "scripts": {
+    "build": "vp pack",
+    "dev": "vp pack --watch",
+    "test": "vp test",
+    "check": "vp check"
+  },
+  "dependencies": {
+    "bailian-cli-core": "workspace:*",
+    "bailian-cli-runtime": "workspace:*",
+    "boxen": "catalog:",
+    "chalk": "catalog:",
+    "yaml": "catalog:"
+  },
+  "devDependencies": {
+    "@types/node": "catalog:",
+    "@typescript/native-preview": "7.0.0-dev.20260328.1",
+    "typescript": "^6.0.2",
+    "vite-plus": "0.1.22"
+  },
+  "engines": {
+    "node": ">=22.12.0"
+  }
+}

packages/commands/src/commands/advisor/recommend.ts

@@ -17,9 +17,9 @@ import {
 } from "bailian-cli-core";
 import boxen from "boxen";
 import chalk, { Chalk, type ChalkInstance } from "chalk";
-import { emitBare, emitResult } from "../../output/output.ts";
-import { createSpinner } from "../../output/progress.ts";
-import { failIfMissing, promptText } from "../../output/prompt.ts";
+import { emitBare, emitResult } from "bailian-cli-runtime";
+import { createSpinner } from "bailian-cli-runtime";
+import { failIfMissing, promptText, cmdUsage } from "bailian-cli-runtime";
 
 function formatContextWindow(tokens: number): string {
   if (tokens >= 1_000_000)
@@ -29,41 +29,41 @@ function formatContextWindow(tokens: number): string {
 }
 
 const MODALITY_LABELS: Record<string, string> = {
-  Text: "文本",
-  Image: "图片",
-  Video: "视频",
-  Audio: "音频",
+  Text: "Text",
+  Image: "Image",
+  Video: "Video",
+  Audio: "Audio",
 };
 const CAPABILITY_LABELS: Record<string, string> = {
-  TG: "文本生成",
-  VU: "视觉理解",
-  IG: "图像生成",
-  VG: "视频生成",
-  TTS: "语音合成",
-  ASR: "语音识别",
-  Reasoning: "推理",
+  TG: "Text Gen",
+  VU: "Vision",
+  IG: "Image Gen",
+  VG: "Video Gen",
+  TTS: "Text-to-Speech",
+  ASR: "Speech-to-Text",
+  Reasoning: "Reasoning",
 };
 const BUDGET_LABELS: Record<string, string> = {
-  low: "低成本优先",
-  medium: "适中",
-  high: "高投入",
+  low: "Cost-Effective",
+  medium: "Balanced",
+  high: "High Investment",
 };
 const QUALITY_LABELS: Record<string, string> = {
-  flagship: "旗舰优先",
-  balanced: "均衡",
-  "cost-optimized": "性价比优先",
+  flagship: "Flagship",
+  balanced: "Balanced",
+  "cost-optimized": "Value",
 };
 const PREFERENCE_MODE_LABELS: Record<string, string> = {
-  scoped: "限定范围",
-  comparison: "对比评估",
-  alternative: "替代推荐",
+  scoped: "Scoped",
+  comparison: "Comparison",
+  alternative: "Alternative",
 };
 
 function formatIntentSummary(intent: IntentProfile, noColor: boolean): string {
   const colorize = noColor ? new Chalk({ level: 0 }) : chalk;
 
   const lines: string[] = [];
-  lines.push(colorize.cyan.bold("需求理解"));
+  lines.push(colorize.cyan.bold("Intent Analysis"));
 
   if (intent.taskSummary) {
     lines.push("");
@@ -72,7 +72,7 @@ function formatIntentSummary(intent: IntentProfile, noColor: boolean): string {
 
   if (intent.scenarioHints.length) {
     lines.push("");
-    lines.push(`${colorize.dim("场景特征")}  ${intent.scenarioHints.join(" · ")}`);
+    lines.push(`${colorize.dim("Scenario")}  ${intent.scenarioHints.join(" · ")}`);
   }
 
   const inputLabels = intent.inputModality.map((mod) => MODALITY_LABELS[mod] ?? mod);
@@ -80,40 +80,40 @@ function formatIntentSummary(intent: IntentProfile, noColor: boolean): string {
   if (inputLabels.length || outputLabels.length) {
     lines.push("");
     const parts: string[] = [];
-    if (inputLabels.length) parts.push(`${colorize.dim("输入")} ${inputLabels.join(", ")}`);
-    if (outputLabels.length) parts.push(`${colorize.dim("输出")} ${outputLabels.join(", ")}`);
+    if (inputLabels.length) parts.push(`${colorize.dim("Input")} ${inputLabels.join(", ")}`);
+    if (outputLabels.length) parts.push(`${colorize.dim("Output")} ${outputLabels.join(", ")}`);
     lines.push(parts.join("    "));
   }
 
   const capLabels = intent.requiredCapabilities.map((cap) => CAPABILITY_LABELS[cap] ?? cap);
   if (capLabels.length) {
-    lines.push(`${colorize.dim("所需能力")}  ${capLabels.join(", ")}`);
+    lines.push(`${colorize.dim("Capabilities")}  ${capLabels.join(", ")}`);
   }
 
   const budgetLabel = BUDGET_LABELS[intent.budget] ?? intent.budget;
   const qualityLabel = QUALITY_LABELS[intent.qualityPreference] ?? intent.qualityPreference;
   lines.push("");
   lines.push(
-    `${colorize.dim("预算倾向")} ${budgetLabel}    ${colorize.dim("质量偏好")} ${qualityLabel}`,
+    `${colorize.dim("Budget")} ${budgetLabel}    ${colorize.dim("Quality")} ${qualityLabel}`,
   );
 
   const preference = intent.modelPreference;
   if (preference && preference.mode !== "unconstrained") {
     lines.push("");
     const modeLabel = PREFERENCE_MODE_LABELS[preference.mode] ?? preference.mode;
-    const prefParts = [colorize.dim("推荐模式") + ` ${colorize.yellow(modeLabel)}`];
+    const prefParts = [colorize.dim("Mode") + ` ${colorize.yellow(modeLabel)}`];
     if (preference.targets?.length) {
-      prefParts.push(colorize.dim("目标") + ` ${preference.targets.join(", ")}`);
+      prefParts.push(colorize.dim("Targets") + ` ${preference.targets.join(", ")}`);
     }
     if (preference.excludes?.length) {
-      prefParts.push(colorize.dim("排除") + ` ${preference.excludes.join(", ")}`);
+      prefParts.push(colorize.dim("Excludes") + ` ${preference.excludes.join(", ")}`);
     }
     lines.push(prefParts.join("    "));
   }
 
   if (intent.segments?.length) {
     lines.push("");
-    lines.push(colorize.dim("任务拆解"));
+    lines.push(colorize.dim("Pipeline"));
     for (const [idx, segment] of intent.segments.entries()) {
       const outMods = segment.outputModality.map((mod) => MODALITY_LABELS[mod] ?? mod).join(", ");
       lines.push(
@@ -131,19 +131,19 @@ function formatIntentSummary(intent: IntentProfile, noColor: boolean): string {
   });
 }
 
-const RECOMMEND_LABELS = ["最佳推荐", "次优选择", "备选参考"];
+const RECOMMEND_LABELS = ["Best Pick", "Runner-Up", "Alternative"];
 
 function renderCard(rec: RecommendedModel, index: number, colorize: ChalkInstance): string {
   const labelColors = [colorize.green.bold, colorize.blue.bold, colorize.magenta.bold];
   const colorFn = labelColors[index] ?? colorize.white.bold;
-  const label = RECOMMEND_LABELS[index] ?? `推荐 #${index + 1}`;
+  const label = RECOMMEND_LABELS[index] ?? `#${index + 1}`;
 
   const lines: string[] = [];
-  lines.push(colorFn(`⬢ 推荐 #${index + 1} — ${label}`));
+  lines.push(colorFn(`⬢  #${index + 1} — ${label}`));
   lines.push("");
   lines.push(`${colorize.bold(rec.name)}  ${colorize.dim(`(${rec.model})`)}`);
   lines.push("");
-  lines.push(`${colorize.cyan("推荐理由")}  ${rec.reason}`);
+  lines.push(`${colorize.cyan("Why")}  ${rec.reason}`);
 
   if (rec.highlights.length) {
     lines.push("");
@@ -153,8 +153,8 @@ function renderCard(rec: RecommendedModel, index: number, colorize: ChalkInstanc
   }
 
   const meta: string[] = [];
-  if (rec.contextWindow) meta.push(`上下文 ${formatContextWindow(rec.contextWindow)}`);
-  if (rec.maxOutputTokens) meta.push(`最大输出 ${formatContextWindow(rec.maxOutputTokens)}`);
+  if (rec.contextWindow) meta.push(`Context ${formatContextWindow(rec.contextWindow)}`);
+  if (rec.maxOutputTokens) meta.push(`Max Output ${formatContextWindow(rec.maxOutputTokens)}`);
   if (meta.length) {
     lines.push("");
     lines.push(colorize.dim(meta.join("  ·  ")));
@@ -163,7 +163,7 @@ function renderCard(rec: RecommendedModel, index: number, colorize: ChalkInstanc
   const docLink = buildDocLink(rec.docUrl);
   if (docLink) {
     lines.push("");
-    lines.push(colorize.dim(`文档  ${docLink}`));
+    lines.push(colorize.dim(`Docs  ${docLink}`));
   }
 
   return boxen(lines.join("\n"), {
@@ -183,7 +183,7 @@ function formatSingleResult(results: RecommendedModel[], noColor: boolean): stri
 function formatPipelineResult(summary: string, steps: PipelineStep[], noColor: boolean): string {
   const colorize = noColor ? new Chalk({ level: 0 }) : chalk;
   const lines: string[] = [];
-  lines.push(`  ${colorize.yellow.bold("⚡ 组合方案")}  ${summary}`);
+  lines.push(`  ${colorize.yellow.bold("⚡ Pipeline")}  ${summary}`);
 
   for (const [stepIdx, { step, recommendations, warnings }] of steps.entries()) {
     lines.push("");
@@ -215,10 +215,9 @@ function isEmptyResult(result: RecommendResult): boolean {
 }
 
 export default defineCommand({
-  name: "advisor recommend",
   description:
     "Recommend the best models for your use case (intent analysis → candidate recall → LLM ranking)",
-  usage: "bl advisor recommend <prompt> [flags]",
+  usageArgs: "<prompt> [flags]",
   options: [
     {
       flag: "--message <text>",
@@ -233,13 +232,13 @@ export default defineCommand({
       description: "Output format: text (default in TTY), json, yaml",
     },
   ],
-  examples: [
-    'bl advisor recommend --message "我要做一个能理解图片的客服机器人"',
-    'bl advisor recommend --message "做一个Agent自动根据用户意图生成动画片"',
-    'bl advisor recommend --message "法律合同审查，要求高精准度"',
-    'bl advisor recommend --message "做一个低成本高并发的在线客服" --output json',
-    'bl advisor recommend --message "长文本摘要" --dry-run',
-    "bl advisor recommend                                          # 交互式输入需求",
+  exampleArgs: [
+    '--message "I need a visual-understanding chatbot"',
+    '--message "Build an Agent that auto-generates animations"',
+    '--message "Legal contract review, high precision required"',
+    '--message "Low-cost high-concurrency online customer service" --output json',
+    '--message "Long document summarization" --dry-run',
+    "                                          # Interactive input",
   ],
   async run(config: Config, flags: GlobalFlags) {
     const positional = ((flags as Record<string, unknown>)._positional as string[]) ?? [];
@@ -247,14 +246,14 @@ export default defineCommand({
 
     if (!userInput.trim()) {
       if (isInteractive({ nonInteractive: config.nonInteractive })) {
-        const hint = await promptText({ message: "描述你的需求:" });
+        const hint = await promptText({ message: "Describe your requirement:" });
         if (!hint) {
-          process.stderr.write("已取消。\n");
+          process.stderr.write("Cancelled.\n");
           process.exit(1);
         }
         userInput = hint;
       } else {
-        failIfMissing("message", 'bl advisor recommend "你的需求"');
+        failIfMissing("message", cmdUsage(config, '"your requirement"'));
       }
     }
 
@@ -262,16 +261,16 @@ export default defineCommand({
     const format = detectOutputFormat(config.output);
 
     const modelsOptions: GetModelsOptions = {
-      onPrepareStart: () => process.stderr.write("初始化中...\n"),
+      onPrepareStart: () => process.stderr.write("Initializing model data...\n"),
     };
-    process.stderr.write("正在分析需求...\n");
+    process.stderr.write("Analyzing your request...\n");
     const [allModels, intent] = await Promise.all([
       getModels(config, modelsOptions),
       analyzeIntent(config, userInput),
     ]);
 
     if (intent.confidence === 0) {
-      process.stderr.write("需求分析超时，使用默认参数继续...\n");
+      process.stderr.write("Intent analysis timed out, using defaults...\n");
     } else {
       process.stderr.write("\n");
     }
@@ -297,7 +296,7 @@ export default defineCommand({
     }
 
     // Stage 3: LLM Ranking
-    const spinner = createSpinner("正在推荐最佳模型...");
+    const spinner = createSpinner("Recommending best models...");
     spinner.start();
 
     const result = await rankModels(config, candidates, intent, userInput, top);
@@ -305,12 +304,31 @@ export default defineCommand({
     spinner.stop();
 
     if (isEmptyResult(result)) {
-      emitBare("暂无满足该需求的模型。");
+      emitBare("No suitable models found for this request.");
       return;
     }
 
     if (format !== "text") {
-      emitResult(result, format);
+      emitResult(
+        {
+          intent: {
+            taskSummary: intent.taskSummary,
+            scenarioHints: intent.scenarioHints,
+            complexity: intent.complexity,
+            inputModality: intent.inputModality,
+            outputModality: intent.outputModality,
+            requiredCapabilities: intent.requiredCapabilities,
+            budget: intent.budget,
+            qualityPreference: intent.qualityPreference,
+            modelPreference:
+              intent.modelPreference?.mode !== "unconstrained" ? intent.modelPreference : undefined,
+            segments: intent.segments,
+          },
+          result,
+          candidates: candidates.length,
+        },
+        format,
+      );
       return;
     }
 

packages/commands/src/commands/app/call.ts

@@ -11,13 +11,12 @@ import {
   type AppStreamChunk,
   type AppCompletionResponse,
 } from "bailian-cli-core";
-import { failIfMissing } from "../../output/prompt.ts";
-import { emitResult, emitBare } from "../../output/output.ts";
+import { failIfMissing, cmdUsage } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
 
 export default defineCommand({
-  name: "app call",
   description: "Call a Bailian application (agent or workflow)",
-  usage: "bl app call --app-id <id> --prompt <text> [flags]",
+  usageArgs: "--app-id <id> --prompt <text> [flags]",
   options: [
     { flag: "--app-id <id>", description: "Application ID (required)", required: true },
     { flag: "--prompt <text>", description: "Input prompt text", required: true },
@@ -34,20 +33,20 @@ export default defineCommand({
     { flag: "--biz-params <json>", description: "Business parameters JSON (workflow variables)" },
     { flag: "--has-thoughts", description: "Show agent thinking process" },
   ],
-  examples: [
-    'bl app call --app-id abc123 --prompt "你好"',
-    'bl app call --app-id abc123 --prompt "描述这张图片" --image https://example.com/photo.jpg',
-    'bl app call --app-id abc123 --prompt "分析图片" --image img1.jpg --image img2.jpg',
-    'bl app call --app-id abc123 --prompt "继续" --session-id sess_xxx --stream',
-    'bl app call --app-id abc123 --prompt "搜索资料" --pipeline-ids pipe1,pipe2',
-    'bl app call --app-id abc123 --prompt "开始" --biz-params \'{"key":"value"}\'',
+  exampleArgs: [
+    '--app-id abc123 --prompt "Hello"',
+    '--app-id abc123 --prompt "Describe this image" --image https://example.com/photo.jpg',
+    '--app-id abc123 --prompt "Analyze the image" --image img1.jpg --image img2.jpg',
+    '--app-id abc123 --prompt "Continue" --session-id sess_xxx --stream',
+    '--app-id abc123 --prompt "Search for materials" --pipeline-ids pipe1,pipe2',
+    '--app-id abc123 --prompt "Start" --biz-params \'{"key":"value"}\'',
   ],
   async run(config: Config, flags: GlobalFlags) {
     const appId = flags.appId as string;
-    if (!appId) failIfMissing("app-id", "bl app call --app-id <id> --prompt <text>");
+    if (!appId) failIfMissing("app-id", cmdUsage(config, "--app-id <id> --prompt <text>"));
 
     const prompt = flags.prompt as string;
-    if (!prompt) failIfMissing("prompt", "bl app call --app-id <id> --prompt <text>");
+    if (!prompt) failIfMissing("prompt", cmdUsage(config, "--app-id <id> --prompt <text>"));
 
     const shouldStream =
       flags.stream === true || (flags.stream === undefined && process.stdout.isTTY);

packages/commands/src/commands/app/list.ts

@@ -6,14 +6,14 @@ import {
   type Config,
   type GlobalFlags,
 } from "bailian-cli-core";
-import { emitResult } from "../../output/output.ts";
+import { emitResult } from "bailian-cli-runtime";
 
 const APP_LIST_API = "zeldaEasy.broadscope-bailian.app-control.list";
 
 export default defineCommand({
-  name: "app list",
   description: "List Bailian applications",
-  usage: "bl app list [flags]",
+  skipDefaultApiKeySetup: true,
+  usageArgs: "[flags]",
   options: [
     {
       flag: "--name <name>",
@@ -29,22 +29,22 @@ export default defineCommand({
       description: "Results per page (default: 30)",
       type: "number",
     },
+    { flag: "--console-region <region>", description: "Console region" },
     {
-      flag: "--region <region>",
-      description: "API region (default: cn-beijing)",
+      flag: "--console-site <site>",
+      description: "Console site: domestic, international",
+    },
+    {
+      flag: "--console-switch-agent <uid>",
+      description: "Switch agent UID",
+      type: "number",
     },
   ],
-  examples: [
-    "bl app list",
-    "bl app list --name 客服",
-    "bl app list --page 2 --page-size 10",
-    "bl app list --output json",
-  ],
+  exampleArgs: ["", "--name customer service", "--page 2 --page-size 10", "--output json"],
   async run(config: Config, flags: GlobalFlags) {
     const name = (flags.name as string) || "";
     const pageNo = (flags.page as number) || 1;
     const pageSize = (flags.pageSize as number) || 30;
-    const region = (flags.region as string) || "cn-beijing";
     const format = detectOutputFormat(config.output);
 
     const credential = await resolveConsoleGatewayCredential(config);
@@ -61,17 +61,13 @@ export default defineCommand({
     };
 
     if (config.dryRun) {
-      emitResult(
-        { api: APP_LIST_API, data, region, token: credential.token.slice(0, 8) + "..." },
-        format,
-      );
+      emitResult({ api: APP_LIST_API, data, token: credential.token.slice(0, 8) + "..." }, format);
       return;
     }
 
     const result = (await callConsoleGateway(config, credential.token, {
       api: APP_LIST_API,
       data,
-      region,
     })) as any;
 
     const list: unknown[] = result?.data?.DataV2?.data?.data?.list ?? [];

packages/commands/src/commands/auth/login-console.ts

@@ -5,18 +5,24 @@ import http from "node:http";
 import {
   BailianError,
   ExitCode,
+  chatEndpoint,
   getConfigPath,
   readConfigFile,
+  requestJson,
   writeConfigFile,
+  type Config,
 } from "bailian-cli-core";
 
 const CONSOLE_LOGIN_TIMEOUT_MS = 15 * 60 * 1000;
 const MAX_AUTH_CALLBACK_BODY = 65536;
 
-const DEFAULT_CONSOLE_ORIGIN = "https://bailian.console.aliyun.com";
+const CONSOLE_ORIGINS: Record<string, string> = {
+  domestic: "https://bailian.console.aliyun.com",
+  international: "https://modelstudio.console.alibabacloud.com",
+};
 
-export function resolveConsoleOrigin(): string {
-  return process.env.BAILIAN_CONSOLE_ORIGIN || DEFAULT_CONSOLE_ORIGIN;
+export function resolveConsoleOrigin(site?: string): string {
+  return (site && CONSOLE_ORIGINS[site]) || CONSOLE_ORIGINS.domestic!;
 }
 
 function readBodyBounded(req: http.IncomingMessage): Promise<string> {
@@ -210,9 +216,76 @@ function parseApiKeyFromRawBody(raw: string, contentType: string): string | null
   return null;
 }
 
+type CallbackExtras = Pick<
+  CallbackCredentials,
+  "baseUrl" | "consoleSite" | "consoleRegion" | "consoleSwitchAgent" | "workspaceId"
+>;
+
+function stringField(o: Record<string, unknown>, ...keys: string[]): string | null {
+  for (const k of keys) {
+    const v = o[k];
+    if (typeof v === "string" && v.trim()) return v.trim();
+  }
+  return null;
+}
+
+function parseExtrasFromRawBody(raw: string, contentType: string): CallbackExtras {
+  const empty: CallbackExtras = {
+    baseUrl: null,
+    consoleSite: null,
+    consoleRegion: null,
+    consoleSwitchAgent: null,
+    workspaceId: null,
+  };
+  if (!raw.trim()) return empty;
+
+  let obj: Record<string, unknown> | null = null;
+
+  const ct = contentType.toLowerCase();
+  if (ct.includes("application/json") || ct.includes("text/json")) {
+    try {
+      const parsed = JSON.parse(raw.trim());
+      if (parsed && typeof parsed === "object" && !Array.isArray(parsed)) obj = parsed;
+    } catch {
+      /* */
+    }
+  }
+  if (!obj && ct.includes("application/x-www-form-urlencoded")) {
+    try {
+      const params = new URLSearchParams(raw.trim());
+      obj = Object.fromEntries(params);
+    } catch {
+      /* */
+    }
+  }
+  if (!obj) {
+    try {
+      const parsed = JSON.parse(raw.trim());
+      if (parsed && typeof parsed === "object" && !Array.isArray(parsed)) obj = parsed;
+    } catch {
+      /* */
+    }
+  }
+
+  if (!obj) return empty;
+
+  return {
+    baseUrl: stringField(obj, "base_url", "baseUrl"),
+    consoleSite: stringField(obj, "console_site", "consoleSite"),
+    consoleRegion: stringField(obj, "console_region", "consoleRegion"),
+    consoleSwitchAgent: stringField(obj, "console_switch_agent", "consoleSwitchAgent"),
+    workspaceId: stringField(obj, "workspace_id", "workspaceId"),
+  };
+}
+
 interface CallbackCredentials {
   accessToken: string | null;
   apiKey: string | null;
+  baseUrl: string | null;
+  consoleSite: string | null;
+  consoleRegion: string | null;
+  consoleSwitchAgent: string | null;
+  workspaceId: string | null;
 }
 
 async function extractCredentialsFromRequest(
@@ -222,12 +295,30 @@ async function extractCredentialsFromRequest(
   const accessTokenFromQuery =
     u.searchParams.get("access_token") ?? u.searchParams.get("accessToken");
   const apiKeyFromQuery = u.searchParams.get("api_key") ?? u.searchParams.get("apiKey");
+  const baseUrlFromQuery = u.searchParams.get("base_url") ?? u.searchParams.get("baseUrl");
+  const consoleSiteFromQuery =
+    u.searchParams.get("console_site") ?? u.searchParams.get("consoleSite");
+  const consoleRegionFromQuery =
+    u.searchParams.get("console_region") ?? u.searchParams.get("consoleRegion");
+  const consoleSwitchAgentFromQuery =
+    u.searchParams.get("console_switch_agent") ?? u.searchParams.get("consoleSwitchAgent");
+  const workspaceIdFromQuery =
+    u.searchParams.get("workspace_id") ?? u.searchParams.get("workspaceId");
+
+  const extras = {
+    baseUrl: baseUrlFromQuery?.trim() || null,
+    consoleSite: consoleSiteFromQuery?.trim() || null,
+    consoleRegion: consoleRegionFromQuery?.trim() || null,
+    consoleSwitchAgent: consoleSwitchAgentFromQuery?.trim() || null,
+    workspaceId: workspaceIdFromQuery?.trim() || null,
+  };
 
   const m = req.method ?? "GET";
   if (m !== "POST" && m !== "PUT" && m !== "PATCH") {
     return {
       accessToken: accessTokenFromQuery?.trim() || null,
       apiKey: apiKeyFromQuery?.trim() || null,
+      ...extras,
     };
   }
 
@@ -239,12 +330,24 @@ async function extractCredentialsFromRequest(
     return {
       accessToken: accessTokenFromQuery?.trim() || null,
       apiKey: apiKeyFromQuery?.trim() || null,
+      ...extras,
     };
   }
 
   const accessToken = accessTokenFromQuery?.trim() || parseAccessTokenFromRawBody(raw, contentType);
   const apiKey = apiKeyFromQuery?.trim() || parseApiKeyFromRawBody(raw, contentType);
-  return { accessToken, apiKey };
+
+  const bodyExtras = parseExtrasFromRawBody(raw, contentType);
+
+  return {
+    accessToken,
+    apiKey,
+    baseUrl: extras.baseUrl || bodyExtras.baseUrl,
+    consoleSite: extras.consoleSite || bodyExtras.consoleSite,
+    consoleRegion: extras.consoleRegion || bodyExtras.consoleRegion,
+    consoleSwitchAgent: extras.consoleSwitchAgent || bodyExtras.consoleSwitchAgent,
+    workspaceId: extras.workspaceId || bodyExtras.workspaceId,
+  };
 }
 
 function listenServerOnFreeLocalPort(server: http.Server): Promise<number> {
@@ -276,9 +379,69 @@ function openInBrowser(url: string): Promise<void> {
   });
 }
 
+const RETRY_DELAY_BASE_MS = 500;
+
+function canRetry(err: unknown): boolean {
+  if (err instanceof BailianError) {
+    if (err.exitCode === ExitCode.NETWORK || err.exitCode === ExitCode.TIMEOUT) return true;
+    const status = err.api?.httpStatus;
+    return status === 401 || (status !== undefined && status >= 500);
+  }
+  if (err instanceof Error) {
+    return (
+      err.name === "AbortError" ||
+      err.name === "TimeoutError" ||
+      err.message.includes("timed out") ||
+      err.message === "fetch failed"
+    );
+  }
+  return false;
+}
+
+export async function validateAndPersistApiKey(
+  config: Config,
+  key: string,
+  baseUrl: string,
+): Promise<void> {
+  process.stderr.write("Testing key... ");
+  const testConfig = { ...config, apiKey: key, baseUrl };
+  const requestOpts = {
+    url: chatEndpoint(testConfig.baseUrl),
+    method: "POST",
+    timeout: Math.min(config.timeout, 30),
+    body: {
+      model: "qwen3.7-max",
+      messages: [{ role: "user", content: "hi" }],
+      max_tokens: 1,
+    },
+  };
+
+  for (let attempt = 1; attempt <= 3; attempt++) {
+    try {
+      await requestJson<unknown>(testConfig, requestOpts);
+      break;
+    } catch (err) {
+      if (attempt >= 3 || !canRetry(err)) {
+        process.stderr.write("Failed\n");
+        throw new BailianError("API key validation failed", ExitCode.AUTH, "Invalid API key.", {
+          cause: err,
+        });
+      }
+      const delayMs = RETRY_DELAY_BASE_MS * 2 ** (attempt - 1);
+      await new Promise((resolve) => setTimeout(resolve, delayMs));
+    }
+  }
+
+  process.stderr.write("Valid\n");
+  const existing = readConfigFile() as Record<string, unknown>;
+  existing.api_key = key;
+  await writeConfigFile(existing);
+}
+
 export async function runConsoleLogin(
   consoleOrigin: string,
-  opts?: { needApiKey?: boolean; onApiKey?: (key: string) => Promise<void> },
+  config: Config,
+  opts?: { needApiKey?: boolean },
 ): Promise<void> {
   const state = randomBytes(16).toString("hex");
   let callbackError: unknown;
@@ -301,18 +464,35 @@ export async function runConsoleLogin(
         return;
       }
 
-      const { accessToken, apiKey } = await extractCredentialsFromRequest(req);
+      const {
+        accessToken,
+        apiKey,
+        baseUrl,
+        consoleSite,
+        consoleRegion,
+        consoleSwitchAgent,
+        workspaceId,
+      } = await extractCredentialsFromRequest(req);
 
-      if (accessToken || apiKey) {
+      const hasConfig =
+        accessToken || baseUrl || consoleSite || consoleRegion || consoleSwitchAgent || workspaceId;
+
+      if (hasConfig || apiKey) {
         try {
-          if (accessToken) {
+          if (hasConfig) {
             const existing = readConfigFile() as Record<string, unknown>;
-            existing.access_token = accessToken;
+            if (accessToken) existing.access_token = accessToken;
+            if (baseUrl) existing.base_url = baseUrl;
+            if (consoleSite) existing.console_site = consoleSite;
+            if (consoleRegion) existing.console_region = consoleRegion;
+            if (consoleSwitchAgent) existing.console_switch_agent = Number(consoleSwitchAgent);
+            if (workspaceId) existing.workspace_id = workspaceId;
             await writeConfigFile(existing);
-            process.stderr.write(`access_token saved to ${getConfigPath()}\n`);
+            process.stderr.write(`Config saved to ${getConfigPath()}\n`);
           }
-          if (apiKey && opts?.onApiKey) {
-            await opts.onApiKey(apiKey);
+          if (apiKey) {
+            const testBaseUrl = baseUrl || config.baseUrl;
+            await validateAndPersistApiKey(config, apiKey, testBaseUrl);
           }
         } catch (err: unknown) {
           callbackError = err;
@@ -329,7 +509,7 @@ export async function runConsoleLogin(
       });
       res.end("OK\n");
 
-      if (accessToken || apiKey) {
+      if (hasConfig || apiKey) {
         server.close();
       }
     } catch {

packages/commands/src/commands/auth/login.ts

@@ -0,0 +1,90 @@
+import {
+  defineCommand,
+  isInteractive,
+  maskToken,
+  readConfigFile,
+  writeConfigFile,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { printQuickStart } from "bailian-cli-runtime";
+import { emitBare } from "bailian-cli-runtime";
+import { promptConfirm } from "bailian-cli-runtime";
+import { printCurrentCommandHelp } from "bailian-cli-runtime";
+import {
+  resolveConsoleOrigin,
+  runConsoleLogin,
+  validateAndPersistApiKey,
+} from "./login-console.ts";
+
+export default defineCommand({
+  description: "Authenticate with API key or console browser login (credentials can coexist)",
+  skipDefaultApiKeySetup: true,
+  usageArgs: "--api-key <key> | --console",
+  options: [
+    { flag: "--api-key <key>", description: "DashScope API key to store" },
+    {
+      flag: "--base-url <url>",
+      description: "DashScope API base URL (used with --api-key for validation)",
+    },
+    {
+      flag: "--console",
+      description:
+        "Sign in via browser; use --console-site to choose domestic (default) or international",
+    },
+  ],
+  exampleArgs: ["--api-key sk-xxxxx", "--console"],
+  async run(config: Config, flags: GlobalFlags) {
+    if (flags.console) {
+      if (config.dryRun) {
+        emitBare(
+          "Would bind a free port on 127.0.0.1 and open the console login URL in your browser.",
+        );
+        return;
+      }
+      const hasApiKey = !!(config.apiKey || config.fileApiKey);
+      await runConsoleLogin(resolveConsoleOrigin(config.consoleSite || "domestic"), config, {
+        needApiKey: !hasApiKey,
+      });
+      return;
+    }
+
+    const envKey = process.env.DASHSCOPE_API_KEY;
+    if (envKey && !flags.apiKey) {
+      const maskedEnvKey = maskToken(envKey);
+      if (isInteractive({ nonInteractive: config.nonInteractive })) {
+        const proceed = await promptConfirm({
+          message: `Detected DASHSCOPE_API_KEY in environment (${maskedEnvKey}).\nYou are already authenticated via env.\nDo you still want to configure local persistent credentials?`,
+          initialValue: false,
+        });
+        if (!proceed) {
+          process.stdout.write("Login skipped. Using environment variables.\n");
+          process.exit(0);
+        }
+      } else {
+        process.stderr.write(`Warning: DASHSCOPE_API_KEY is already set in environment.\n`);
+      }
+    }
+
+    const key = (flags.apiKey as string) || config.apiKey;
+    if (!key) {
+      printCurrentCommandHelp(process.stderr);
+      process.exit(0);
+    }
+
+    const baseUrl = (flags.baseUrl as string) || undefined;
+    const effectiveConfig = baseUrl ? { ...config, baseUrl } : config;
+
+    if (!config.dryRun) {
+      if (baseUrl) {
+        const existing = readConfigFile() as Record<string, unknown>;
+        existing.base_url = baseUrl;
+        await writeConfigFile(existing);
+      }
+      await validateAndPersistApiKey(effectiveConfig, key, effectiveConfig.baseUrl);
+      printQuickStart();
+    } else {
+      emitBare("Would validate and save API key.");
+    }
+  },
+});

packages/commands/src/commands/auth/logout.ts

@@ -7,7 +7,7 @@ import {
   type Config,
   type GlobalFlags,
 } from "bailian-cli-core";
-import { emitBare } from "../../output/output.ts";
+import { emitBare } from "bailian-cli-runtime";
 
 async function clearConsoleToken(): Promise<boolean> {
   const file = readConfigFile() as Record<string, unknown>;
@@ -18,9 +18,9 @@ async function clearConsoleToken(): Promise<boolean> {
 }
 
 export default defineCommand({
-  name: "auth logout",
   description: "Clear stored credentials",
-  usage: "bl auth logout [--console] [--yes] [--dry-run]",
+  skipDefaultApiKeySetup: true,
+  usageArgs: "[--console] [--yes] [--dry-run]",
   options: [
     {
       flag: "--console",
@@ -29,12 +29,7 @@ export default defineCommand({
     },
     { flag: "--yes", description: "Skip confirmation prompt" },
   ],
-  examples: [
-    "bl auth logout",
-    "bl auth logout --console",
-    "bl auth logout --dry-run",
-    "bl auth logout --yes",
-  ],
+  exampleArgs: ["", "--console", "--dry-run", "--yes"],
   async run(config: Config, flags: GlobalFlags) {
     const file = readConfigFile();
 

packages/commands/src/commands/auth/status.ts

@@ -8,8 +8,8 @@ import {
   type GlobalFlags,
   type ResolvedCredential,
 } from "bailian-cli-core";
-import { emitResult, emitBare } from "../../output/output.ts";
-import { API_KEY_PAGE } from "../../urls.ts";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+import { API_KEY_PAGE } from "bailian-cli-runtime";
 
 interface StoredCredential {
   configured: boolean;
@@ -108,7 +108,7 @@ function hasAnyAuth(status: AuthStatusPayload): boolean {
   );
 }
 
-function emitTextStatus(status: AuthStatusPayload): void {
+function emitTextStatus(status: AuthStatusPayload, config: Config): void {
   emitBare("Authentication Status:");
   emitBare("  Stored credentials (can coexist):");
   if (status.api_key.configured) {
@@ -134,15 +134,25 @@ function emitTextStatus(status: AuthStatusPayload): void {
       `    Console gateway: ${status.console_gateway_commands.method} (${status.console_gateway_commands.source})  ${status.console_gateway_commands.masked}`,
     );
   } else {
-    emitBare("    Console gateway: unavailable (run bl auth login --console)");
+    emitBare(`    Console gateway: unavailable (run ${config.binName} auth login --console)`);
   }
 }
 
 export default defineCommand({
-  name: "auth status",
   description: "Show current authentication state",
-  usage: "bl auth status",
-  examples: ["bl auth status", "bl auth status --output json"],
+  options: [
+    { flag: "--console-region <region>", description: "Console region" },
+    {
+      flag: "--console-site <site>",
+      description: "Console site: domestic, international",
+    },
+    {
+      flag: "--console-switch-agent <uid>",
+      description: "Switch agent UID",
+      type: "number",
+    },
+  ],
+  exampleArgs: ["", "--output json"],
   async run(config: Config, _flags: GlobalFlags) {
     const format = detectOutputFormat(config.output);
     const status = await buildStatus(config);
@@ -152,8 +162,8 @@ export default defineCommand({
         authenticated: false,
         message: "Not authenticated.",
         hint: [
-          "DashScope API: bl auth login --api-key <key> or DASHSCOPE_API_KEY",
-          "Console gateway: bl auth login --console or DASHSCOPE_ACCESS_TOKEN",
+          `DashScope API: ${config.binName} auth login --api-key <key> or DASHSCOPE_API_KEY`,
+          `Console gateway: ${config.binName} auth login --console or DASHSCOPE_ACCESS_TOKEN`,
           `Get API Key: ${API_KEY_PAGE}`,
         ].join("\n"),
         ...status,
@@ -167,6 +177,6 @@ export default defineCommand({
       return;
     }
 
-    emitTextStatus(status);
+    emitTextStatus(status, config);
   },
 });

packages/commands/src/commands/config/set.ts

@@ -9,10 +9,9 @@ import {
   type GlobalFlags,
   ExitCode,
 } from "bailian-cli-core";
-import { emitResult } from "../../output/output.ts";
+import { emitResult, cmdUsage } from "bailian-cli-runtime";
 
 const VALID_KEYS = [
-  "region",
   "base_url",
   "output",
   "output_dir",
@@ -51,21 +50,21 @@ const KEY_ALIASES: Record<string, string> = {
 };
 
 export default defineCommand({
-  name: "config set",
   description: "Set a config value",
-  usage: "bl config set --key <key> --value <value>",
+  skipDefaultApiKeySetup: true,
+  usageArgs: "--key <key> --value <value>",
   options: [
     {
       flag: "--key <key>",
       description:
-        "Config key (region, base_url, output, output_dir, timeout, api_key, access_token, default_*_model, access_key_id, access_key_secret, workspace_id)",
+        "Config key (base_url, output, output_dir, timeout, api_key, access_token, default_*_model, access_key_id, access_key_secret, workspace_id)",
     },
     { flag: "--value <value>", description: "Value to set" },
   ],
-  examples: [
-    "bl config set --key output --value json",
-    "bl config set --key timeout --value 600",
-    "bl config set --key base_url --value https://dashscope.aliyuncs.com",
+  exampleArgs: [
+    "--key output --value json",
+    "--key timeout --value 600",
+    "--key base_url --value https://dashscope.aliyuncs.com",
   ],
   async run(config: Config, flags: GlobalFlags) {
     const key = flags.key as string | undefined;
@@ -75,7 +74,7 @@ export default defineCommand({
       throw new BailianError(
         "--key and --value are required.",
         ExitCode.USAGE,
-        "bl config set --key <key> --value <value>",
+        cmdUsage(config, "--key <key> --value <value>"),
       );
     }
 
@@ -90,13 +89,6 @@ export default defineCommand({
     }
 
     // Validate specific values
-    if (resolvedKey === "region" && !["cn", "us", "intl"].includes(value)) {
-      throw new BailianError(
-        `Invalid region "${value}". Valid values: cn, us, intl`,
-        ExitCode.USAGE,
-      );
-    }
-
     if (resolvedKey === "output" && !["text", "json"].includes(value)) {
       throw new BailianError(
         `Invalid output format "${value}". Valid values: text, json`,

packages/commands/src/commands/config/show.ts

@@ -0,0 +1,39 @@
+import {
+  defineCommand,
+  readConfigFile as loadConfigFile,
+  getConfigPath,
+  detectOutputFormat,
+  maskToken,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { emitResult } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "Display current configuration",
+  skipDefaultApiKeySetup: true,
+  exampleArgs: ["", "--output json"],
+  async run(config: Config, _flags: GlobalFlags) {
+    const file = loadConfigFile();
+    const format = detectOutputFormat(config.output);
+
+    const result: Record<string, unknown> = {
+      ...file,
+      base_url: config.baseUrl,
+      output: config.output,
+      timeout: config.timeout,
+      config_file: getConfigPath(),
+    };
+
+    if (typeof result.api_key === "string") result.api_key = maskToken(result.api_key);
+    if (typeof result.access_token === "string")
+      result.access_token = maskToken(result.access_token);
+    if (typeof result.access_key_id === "string")
+      result.access_key_id = maskToken(result.access_key_id);
+    if (typeof result.access_key_secret === "string") {
+      result.access_key_secret = maskToken(result.access_key_secret);
+    }
+
+    emitResult(result, format);
+  },
+});

packages/commands/src/commands/console/call.ts

@@ -1,6 +1,7 @@
 import {
   defineCommand,
   callConsoleGateway,
+  effectiveConsoleGatewayConfig,
   resolveConsoleGatewayCredential,
   CONSOLE_GATEWAY_NO_TOKEN_MESSAGE,
   BailianError,
@@ -8,13 +9,13 @@ import {
   type Config,
   type GlobalFlags,
 } from "bailian-cli-core";
-import { failIfMissing } from "../../output/prompt.ts";
-import { emitResult } from "../../output/output.ts";
+import { failIfMissing, cmdUsage } from "bailian-cli-runtime";
+import { emitResult } from "bailian-cli-runtime";
 
 export default defineCommand({
-  name: "console call",
   description: "Call a Bailian console API via the CLI gateway",
-  usage: "bl console call --api <api> --data <json> [flags]",
+  skipDefaultApiKeySetup: true,
+  usageArgs: "--api <api> --data <json> [flags]",
   options: [
     {
       flag: "--api <api>",
@@ -26,21 +27,27 @@ export default defineCommand({
       description: "Request data as JSON string",
       required: true,
     },
+    { flag: "--console-region <region>", description: "Console region" },
     {
-      flag: "--region <region>",
-      description: "API region (default: cn-beijing)",
+      flag: "--console-site <site>",
+      description: "Console site: domestic, international",
+    },
+    {
+      flag: "--console-switch-agent <uid>",
+      description: "Switch agent UID",
+      type: "number",
     },
   ],
-  examples: [
-    `bl console call --api zeldaEasy.broadscope-bailian.freeTrial.queryFreeTierQuota --data '{"queryFreeTierQuotaRequest":{"models":["qwen3-max"]}}'`,
-    `bl console call --api some.api.name --data '{"key":"value"}' --region cn-beijing`,
+  exampleArgs: [
+    `--api zeldaEasy.broadscope-bailian.freeTrial.queryFreeTierQuota --data '{"queryFreeTierQuotaRequest":{"models":["qwen3-max"]}}'`,
+    `--api some.api.name --data '{"key":"value"}' --console-region cn-beijing`,
   ],
   async run(config: Config, flags: GlobalFlags) {
     const api = flags.api as string;
-    if (!api) failIfMissing("api", "bl console call --api <api> --data <json>");
+    if (!api) failIfMissing("api", cmdUsage(config, "--api <api> --data <json>"));
 
     const dataRaw = flags.data as string;
-    if (!dataRaw) failIfMissing("data", "bl console call --api <api> --data <json>");
+    if (!dataRaw) failIfMissing("data", cmdUsage(config, "--api <api> --data <json>"));
 
     let data: Record<string, unknown>;
     try {
@@ -50,7 +57,6 @@ export default defineCommand({
       process.exit(1);
     }
 
-    const region = (flags.region as string) || "cn-beijing";
     const format = detectOutputFormat(config.output);
 
     let token: string | undefined;
@@ -63,14 +69,21 @@ export default defineCommand({
     }
 
     if (config.dryRun) {
-      emitResult({ api, data, region, token: token ? token.slice(0, 8) + "..." : null }, format);
+      emitResult(
+        {
+          api,
+          data,
+          token: token ? token.slice(0, 8) + "..." : null,
+          ...effectiveConsoleGatewayConfig(config),
+        },
+        format,
+      );
       return;
     }
 
     const result = await callConsoleGateway(config, token, {
       api,
       data,
-      region,
     });
 
     emitResult(result, format);

packages/commands/src/commands/dataset/delete.ts

@@ -0,0 +1,61 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  deleteDataset,
+  isInteractive,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing, promptConfirm } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "Delete a dataset file by ID",
+  usageArgs: "--file-id <id> [--yes]",
+  options: [
+    { flag: "--file-id <id>", description: "Dataset file ID (required)", required: true },
+    { flag: "--yes", description: "Skip the confirmation prompt", type: "boolean" },
+  ],
+  exampleArgs: ["--file-id file-id-xxx", "--file-id file-id-xxx --yes"],
+  async run(config: Config, flags: GlobalFlags) {
+    const fileId = flags.fileId as string | undefined;
+    if (!fileId) failIfMissing("file-id", "bl dataset delete --file-id <id>");
+
+    const format = detectOutputFormat(config.output);
+    const yes = Boolean(flags.yes);
+
+    if (config.dryRun) {
+      emitResult({ action: "dataset.delete", file_id: fileId }, format);
+      return;
+    }
+
+    if (!yes) {
+      if (isInteractive({ nonInteractive: config.nonInteractive })) {
+        const ok = await promptConfirm({
+          message: `Permanently delete dataset file ${fileId}? This cannot be undone.`,
+          initialValue: false,
+        });
+        if (!ok) {
+          emitBare("Aborted.");
+          return;
+        }
+      } else {
+        throw new BailianError(
+          `Refusing to delete ${fileId} without --yes in non-interactive mode.`,
+          ExitCode.USAGE,
+          "Pass --yes to skip the confirmation prompt.",
+        );
+      }
+    }
+
+    const response = await deleteDataset(config, fileId!);
+
+    if (config.quiet || format === "text") {
+      emitBare(`Deleted ${fileId}.`);
+    } else {
+      emitResult(response, format);
+    }
+  },
+});

packages/commands/src/commands/dataset/get.ts

@@ -0,0 +1,60 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  getDataset,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "Get details of a single dataset file",
+  usageArgs: "--file-id <id>",
+  options: [{ flag: "--file-id <id>", description: "Dataset file ID (required)", required: true }],
+  exampleArgs: ["--file-id file-xxx", "--file-id file-xxx --output json"],
+  async run(config: Config, flags: GlobalFlags) {
+    const fileId = flags.fileId as string | undefined;
+    if (!fileId) failIfMissing("file-id", "bl dataset get --file-id <id>");
+
+    const format = detectOutputFormat(config.output);
+
+    if (config.dryRun) {
+      emitResult({ action: "dataset.get", file_id: fileId }, format);
+      return;
+    }
+
+    const response = await getDataset(config, fileId!);
+    const file = response.data;
+
+    if (!file) {
+      emitBare(`No data returned for ${fileId}`);
+      return;
+    }
+
+    const sizeKb = file.size !== undefined ? `${(file.size / 1024).toFixed(1)} KB` : "?";
+    const item = {
+      file_id: file.file_id ?? fileId,
+      name: file.name ?? "",
+      size: sizeKb,
+      md5: file.md5 ?? "",
+      purpose: file.purpose ?? "",
+      created_at: file.gmt_create ?? "",
+      description: file.description ?? "",
+    };
+
+    if (format === "json") {
+      emitResult(item, format);
+      return;
+    }
+
+    // text / quiet
+    emitBare(`file_id:      ${item.file_id}`);
+    emitBare(`name:         ${item.name}`);
+    emitBare(`size:         ${item.size}`);
+    if (item.md5) emitBare(`md5:          ${item.md5}`);
+    if (item.purpose) emitBare(`purpose:      ${item.purpose}`);
+    if (item.created_at) emitBare(`created_at:   ${item.created_at}`);
+    if (item.description) emitBare(`description:  ${item.description}`);
+  },
+});

packages/commands/src/commands/dataset/list.ts

@@ -0,0 +1,65 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  listDatasets,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+import { formatTable } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "List uploaded dataset files",
+  usageArgs: "[--page <n>] [--page-size <n>] [--purpose <name>]",
+  options: [
+    { flag: "--page <n>", description: "Page number (default: 1)", type: "number" },
+    {
+      flag: "--page-size <n>",
+      description: "Results per page (default: 10, max 100)",
+      type: "number",
+    },
+    {
+      flag: "--purpose <name>",
+      description: 'Filter by purpose (e.g. "fine-tune", "evaluation"). Omit to list all.',
+    },
+  ],
+  exampleArgs: ["", "--purpose fine-tune", "--purpose evaluation --page-size 20", "--output json"],
+  async run(config: Config, flags: GlobalFlags) {
+    const format = detectOutputFormat(config.output);
+    const pageNo = flags.page !== undefined ? (flags.page as number) : undefined;
+    const pageSize = flags.pageSize !== undefined ? (flags.pageSize as number) : undefined;
+    const purpose = (flags.purpose as string | undefined) || undefined;
+
+    if (config.dryRun) {
+      emitResult({ action: "dataset.list", page: pageNo, page_size: pageSize, purpose }, format);
+      return;
+    }
+
+    const response = await listDatasets(config, { pageNo, pageSize, purpose });
+    const files = response.data?.files ?? [];
+    const total = response.data?.total;
+
+    // Normalize to consistent structure for both text/json output.
+    const items = files.map((item) => ({
+      file_id: item.file_id ?? "",
+      name: item.name ?? "",
+      size: item.size !== undefined ? `${(item.size / 1024).toFixed(1)} KB` : "?",
+      purpose: item.purpose ?? "",
+    }));
+
+    if (format === "json") {
+      emitResult({ items, total }, format);
+      return;
+    }
+
+    // text / quiet
+    if (items.length === 0) {
+      emitBare("No dataset files found.");
+      return;
+    }
+    const headers = ["FILE_ID", "NAME", "SIZE", "PURPOSE"];
+    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}`);
+  },
+});

packages/commands/src/commands/dataset/upload.ts

@@ -0,0 +1,138 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  uploadDataset,
+  validateDataset,
+  parseDatasetSchemaFlag,
+  formatIssue,
+  MAX_DATASET_BYTES,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+  type DatasetFile,
+} from "bailian-cli-core";
+import { failIfMissing } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "Upload a dataset file (.jsonl) to Bailian",
+  usageArgs:
+    "--file <path> [--purpose <name>] [--schema <chatml|dpo|cpt>] [--no-validate] [--full-validate]",
+  options: [
+    {
+      flag: "--file <path>",
+      description: "Local .jsonl dataset file (≤300MB)",
+      required: true,
+    },
+    {
+      flag: "--purpose <name>",
+      description: 'Dataset purpose tag (default: "fine-tune"; e.g. "evaluation")',
+    },
+    {
+      flag: "--schema <s>",
+      description:
+        'Record schema: "chatml" (SFT), "dpo" (chosen/rejected), or "cpt" (raw text). Default auto-detects per record.',
+    },
+    {
+      flag: "--no-validate",
+      description: "Skip the local JSONL pre-flight check (not recommended)",
+      type: "boolean",
+    },
+    {
+      flag: "--full-validate",
+      description: "JSON.parse every line instead of sampling (slower)",
+      type: "boolean",
+    },
+  ],
+  exampleArgs: [
+    "--file train.jsonl",
+    "--file dpo.jsonl --schema dpo",
+    "--file cpt.jsonl --schema cpt",
+    "--file eval.jsonl --purpose evaluation",
+    "--file train.jsonl --full-validate",
+    "--file train.jsonl --no-validate",
+  ],
+  notes: [
+    "Only .jsonl is supported in this release. Three record schemas are",
+    "recognized: chatml = {messages:[...]} (SFT); dpo = {messages:[...],",
+    "chosen, rejected} where chosen/rejected are single assistant messages;",
+    'cpt = {text:"..."} (continual pre-training, raw text). With no --schema,',
+    "a record carrying chosen/rejected is validated as DPO, one with text (and",
+    "no messages) as CPT, otherwise as ChatML. Pass --schema dpo / cpt to",
+    "require that shape on every record, or --schema chatml to ignore the",
+    "preference / text fields. Other purposes may carry a different schema in",
+    "the future and would be served by a purpose-specific validator.",
+    "The dataset upload cap is 300MB per file.",
+    "Upload uses the OpenAI-compatible /compatible-mode/v1/files endpoint so",
+    "the purpose tag is persisted (the DashScope-native /api/v1/files drops it).",
+  ],
+  async run(config: Config, flags: GlobalFlags) {
+    const filePath = flags.file as string | undefined;
+    if (!filePath) failIfMissing("file", "bl dataset upload --file <path>");
+
+    const purpose = (flags.purpose as string | undefined) || "fine-tune";
+    const skipValidate = Boolean(flags.noValidate);
+    const fullValidate = Boolean(flags.fullValidate);
+    const schema = parseDatasetSchemaFlag(flags.schema as string | undefined);
+    const format = detectOutputFormat(config.output);
+
+    if (!skipValidate) {
+      const result = await validateDataset(filePath!, { fullValidate, schema });
+      if (!result.valid) {
+        const lines = [
+          `Dataset validation failed for ${filePath}`,
+          ...result.errors.slice(0, 10).map(formatIssue),
+        ];
+        if (result.errors.length > 10) {
+          lines.push(`  … and ${result.errors.length - 10} more error(s).`);
+        }
+        lines.push(
+          "",
+          "Hint: re-run `bl dataset validate --file <path>` for the full report,",
+          "      or pass --no-validate to skip this check at your own risk.",
+        );
+        throw new BailianError(lines.join("\n"), ExitCode.GENERAL);
+      }
+      // Surface warnings to stderr but keep going.
+      if (result.warnings.length > 0 && !config.quiet) {
+        process.stderr.write(
+          `Dataset validation passed with ${result.warnings.length} warning(s):\n`,
+        );
+        for (const warning of result.warnings.slice(0, 5))
+          process.stderr.write(`${formatIssue(warning)}\n`);
+        if (result.warnings.length > 5) {
+          process.stderr.write(`  … and ${result.warnings.length - 5} more.\n`);
+        }
+      }
+    }
+
+    if (config.dryRun) {
+      emitResult(
+        {
+          action: "dataset.upload",
+          file: filePath,
+          purpose,
+          max_bytes: MAX_DATASET_BYTES,
+          validate: !skipValidate,
+          schema: schema ?? "auto",
+        },
+        format,
+      );
+      return;
+    }
+
+    const uploaded: DatasetFile = await uploadDataset(config, {
+      filePath: filePath!,
+      purpose,
+    });
+
+    if (config.quiet) {
+      emitBare(uploaded.file_id);
+    } else if (format === "text") {
+      emitBare(`Uploaded ${uploaded.name} → file_id=${uploaded.file_id}`);
+    } else {
+      emitResult(uploaded, format);
+    }
+  },
+});

packages/commands/src/commands/dataset/validate.ts

@@ -0,0 +1,120 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  validateDataset,
+  parseDatasetSchemaFlag,
+  formatIssue,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+  type ValidationResult,
+} from "bailian-cli-core";
+import { failIfMissing } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+function formatStats(result: ValidationResult): string[] {
+  const out: string[] = [];
+  if (result.stats.totalRecords !== undefined) out.push(`records: ${result.stats.totalRecords}`);
+  if (result.stats.sampledRecords !== undefined)
+    out.push(`sampled: ${result.stats.sampledRecords}`);
+  if (result.stats.bytes !== undefined) out.push(`bytes: ${result.stats.bytes}`);
+  if (result.stats.durationMs !== undefined) out.push(`took: ${result.stats.durationMs}ms`);
+  return out;
+}
+
+export default defineCommand({
+  description: "Locally validate a dataset file (.jsonl) without uploading",
+  // 纯本地校验，不触网、不需 API key（与 `pipeline validate` 一致）。
+  skipDefaultApiKeySetup: true,
+  usageArgs: "--file <path> [--full-validate] [--schema <chatml|dpo|cpt>]",
+  options: [
+    { flag: "--file <path>", description: "Local .jsonl dataset file", required: true },
+    {
+      flag: "--full-validate",
+      description: "JSON.parse every line instead of sampling (slower)",
+      type: "boolean",
+    },
+    {
+      flag: "--schema <s>",
+      description:
+        'Record schema: "chatml" (SFT), "dpo" (chosen/rejected), or "cpt" (raw text). Default auto-detects per record.',
+    },
+  ],
+  exampleArgs: [
+    "--file train.jsonl",
+    "--file dpo.jsonl --schema dpo",
+    "--file cpt.jsonl --schema cpt",
+    "--file eval.jsonl --full-validate",
+    "--file train.jsonl --output json",
+  ],
+  notes: [
+    "Default scan: every line gets a structural check, then ~160 lines (front 50,",
+    "evenly spaced 100, last 10) are JSON.parsed against the active schema.",
+    "Schemas: chatml = {messages:[...]} (SFT); dpo = {messages:[...], chosen,",
+    "rejected} where chosen/rejected are single assistant messages; cpt =",
+    '{text:"..."} (continual pre-training, raw text). With no --schema, a',
+    "record carrying chosen/rejected is validated as DPO, one with text (and no",
+    "messages) as CPT, otherwise as ChatML. Pass --schema dpo / cpt to require",
+    "that shape on every record (strict), or --schema chatml to ignore the",
+    "preference / text fields. Use --full-validate to JSON.parse every line.",
+  ],
+  async run(config: Config, flags: GlobalFlags) {
+    const filePath = flags.file as string | undefined;
+    if (!filePath) failIfMissing("file", "bl dataset validate --file <path>");
+
+    const fullValidate = Boolean(flags.fullValidate);
+    const schema = parseDatasetSchemaFlag(flags.schema as string | undefined);
+    const format = detectOutputFormat(config.output);
+
+    if (config.dryRun) {
+      emitResult(
+        {
+          action: "dataset.validate",
+          file: filePath,
+          full: fullValidate,
+          schema: schema ?? "auto",
+        },
+        format,
+      );
+      return;
+    }
+
+    const result = await validateDataset(filePath!, { fullValidate, schema });
+
+    if (format === "json") {
+      // For json output we always emit the structured result, exit code conveys validity.
+      emitResult(result, format);
+    } else if (config.quiet) {
+      emitBare(result.valid ? "ok" : "fail");
+    } else {
+      const status = result.valid ? "PASSED" : "FAILED";
+      emitBare(`Dataset validation ${status} for ${result.filePath}`);
+      const stats = formatStats(result);
+      if (stats.length) emitBare(`  ${stats.join(" · ")}`);
+
+      if (result.errors.length) {
+        emitBare(`Errors (${result.errors.length}):`);
+        for (const error of result.errors.slice(0, 20)) emitBare(formatIssue(error));
+        if (result.errors.length > 20) {
+          emitBare(`  … and ${result.errors.length - 20} more.`);
+        }
+      }
+      if (result.warnings.length) {
+        emitBare(`Warnings (${result.warnings.length}):`);
+        for (const warning of result.warnings.slice(0, 10)) emitBare(formatIssue(warning));
+        if (result.warnings.length > 10) {
+          emitBare(`  … and ${result.warnings.length - 10} more.`);
+        }
+      }
+    }
+
+    if (!result.valid) {
+      // Match the upload command's exit-code convention; details already printed.
+      throw new BailianError(
+        `Dataset validation failed: ${result.errors.length} error(s).`,
+        ExitCode.GENERAL,
+      );
+    }
+  },
+});

packages/commands/src/commands/deploy/create.ts

@@ -0,0 +1,168 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  createDeployment,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing, promptConfirm } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+import { pickPlanStrategy } from "./plans.ts";
+
+/**
+ * `bl deploy create` — create a model deployment.
+ *
+ * Plan-specific behaviour (required flags / body assembly / confirm rows /
+ * auto-pick) lives in `plans.ts` (`PlanStrategy` + `STRATEGIES`). This file
+ * only handles the shared envelope: argument parsing, dispatch, dry-run,
+ * confirmation prompt, and result formatting. Adding a new plan = one entry
+ * in the strategy table; nothing here changes.
+ *
+ * `--model` (model identifier) and `--name` (console display name) are required.
+ */
+export default defineCommand({
+  description: "Create a model deployment",
+  usageArgs:
+    "--model <model_name> --name <display_name> [--plan <plan>] [--template-id <id>] [--capacity <n>] [--billing-method <m>] [--input-tpm <n>] [--output-tpm <n>] [--thinking-output-tpm <n>] [--yes]",
+  options: [
+    {
+      flag: "--model <name>",
+      description: "Model name (catalog model or fine-tuned output) (required)",
+      required: true,
+    },
+    {
+      flag: "--name <display_name>",
+      description: "Console display name for the deployment (required)",
+      required: true,
+    },
+    {
+      flag: "--plan <plan>",
+      description: "Billing plan: lora (default, Token-billed) | ptu (Token-billed) | mu",
+    },
+    {
+      flag: "--template-id <id>",
+      description: "Template id (only used by plan=mu; auto-picked if omitted)",
+    },
+    {
+      flag: "--capacity <n>",
+      description:
+        "Resource units (plan=mu only; required by API; defaults to the template's unit)",
+      type: "number",
+    },
+    {
+      flag: "--billing-method <m>",
+      description: 'Billing method (plan=mu only; default "POST_PAY", the only supported value)',
+    },
+    {
+      flag: "--input-tpm <n>",
+      description: "PTU max input tokens/min (required for plan=ptu)",
+      type: "number",
+    },
+    {
+      flag: "--output-tpm <n>",
+      description: "PTU max output tokens/min (required for plan=ptu)",
+      type: "number",
+    },
+    {
+      flag: "--thinking-output-tpm <n>",
+      description: "PTU max thinking-output tokens/min (optional, some models)",
+      type: "number",
+    },
+    { flag: "--yes", description: "Skip the confirmation prompt", type: "boolean" },
+  ],
+  exampleArgs: [
+    "--model my-qwen-sft --name my-sft-test",
+    "--model qwen3.6-flash-2026-04-16 --name my-flash --plan ptu --input-tpm 10000 --output-tpm 1000",
+    "--model qwen3-8b --name my-qwen3-mu --plan mu",
+    "--model qwen3-8b --name my-qwen3 --plan mu --template-id MU1 --capacity 2 --yes",
+  ],
+  notes: [
+    "Plan defaults to `lora` (Token-billed). Pass --plan to override.",
+    "For plan=ptu (Token-billed, provisioned throughput), --input-tpm and",
+    "--output-tpm are required (the platform rejects creation without an",
+    "explicit ptu_capacity despite the doc listing defaults).",
+    "For plan=mu, `capacity`, `billing_method` and `template_id` are required.",
+    "billing_method defaults to POST_PAY (only supported value); template_id",
+    "and capacity are auto-picked from GET /deployments/models when omitted.",
+    "Use `bl deploy models --source base` to inspect available templates.",
+    "After creation, status starts at PENDING and transitions to RUNNING.",
+    "Invoke the deployed model with: bl text chat --model <deployed_model>",
+    "WARNING: --model is overloaded across commands and refers to DIFFERENT",
+    "values. `bl deploy create --model` takes the exported model_name (e.g.",
+    "`qwen3-8b-ft-...`), but the create response also returns a `deployed_model`",
+    "field (the deployment instance id, e.g. `qwen3-8b-5ecb5f068d79`). The",
+    "inference call `bl text chat --model` must use the `deployed_model` from",
+    "the create response — NOT the `model_name` you passed to `deploy create`.",
+    "Do not reuse the value across the two commands.",
+  ],
+  async run(config: Config, flags: GlobalFlags) {
+    const model = flags.model as string | undefined;
+    const name = flags.name as string | undefined;
+    if (!model)
+      failIfMissing("model", "bl deploy create --model <model_name> --name <display_name>");
+    if (!name) failIfMissing("name", "bl deploy create --model <model_name> --name <display_name>");
+
+    const plan = (flags.plan as string | undefined) || "lora";
+    const format = detectOutputFormat(config.output);
+
+    // Plan-specific behaviour is owned by `plans.ts`. The strategy:
+    //   1. Validates required flags (USAGE error if missing).
+    //   2. Resolves the body fragment + confirm rows (mu may auto-pick a
+    //      template from the deployable-models catalog).
+    // Anything outside the strategy table is rejected with a USAGE error.
+    const strategy = pickPlanStrategy(plan);
+    strategy.validateFlags(flags);
+    const resolved = await strategy.resolve({ config, flags, model: model!, name: name! });
+    const body: Record<string, unknown> = {
+      model_name: model!,
+      name: name!,
+      plan,
+      ...resolved.body,
+    };
+
+    if (config.dryRun) {
+      emitResult({ action: "deploy.create", body }, format);
+      return;
+    }
+
+    if (!flags.yes && !config.nonInteractive && !config.quiet) {
+      const lines = [
+        "Create deployment:",
+        `  model:        ${model}`,
+        `  name:         ${name}`,
+        `  plan:         ${plan}${resolved.planLabelSuffix ?? ""}`,
+        ...resolved.confirmRows,
+      ];
+      process.stderr.write(lines.join("\n") + "\n");
+      const ok = await promptConfirm({ message: "Proceed?", initialValue: true });
+      if (!ok) {
+        emitBare("Cancelled.");
+        return;
+      }
+    } else if (!flags.yes && config.nonInteractive) {
+      throw new BailianError(
+        "Pass --yes to confirm deployment creation in non-interactive mode.",
+        ExitCode.USAGE,
+      );
+    }
+
+    const response = await createDeployment(config, body as never);
+    const deployment = response.output ?? response.data;
+
+    if (config.quiet) {
+      emitBare(deployment?.deployed_model ?? "");
+    } else if (format === "text") {
+      emitBare(`Created deployment.`);
+      if (deployment?.deployed_model) emitBare(`  deployed_model:  ${deployment.deployed_model}`);
+      if (deployment?.status) emitBare(`  status:          ${deployment.status}`);
+      if (deployment?.plan) emitBare(`  plan:            ${deployment.plan}`);
+      emitBare(
+        `\nNext: track readiness with: bl deploy get --deployed-model ${deployment?.deployed_model ?? "<id>"}`,
+      );
+    } else {
+      emitResult(response, format);
+    }
+  },
+});

packages/commands/src/commands/deploy/delete.ts

@@ -0,0 +1,93 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  deleteDeployment,
+  getDeployment,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing, promptConfirm } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+/**
+ * `bl deploy delete` — destroy a deployment.
+ *
+ * Server-side precondition: status must be STOPPED or FAILED. We surface a
+ * clear local hint for RUNNING / PENDING deployments before issuing the
+ * DELETE call.
+ */
+export default defineCommand({
+  description: "Delete a model deployment (must be STOPPED or FAILED)",
+  usageArgs: "--deployed-model <id> [--yes] [--skip-precheck]",
+  options: [
+    {
+      flag: "--deployed-model <id>",
+      description: "Deployed model identifier (required)",
+      required: true,
+    },
+    { flag: "--yes", description: "Skip the confirmation prompt", type: "boolean" },
+    {
+      flag: "--skip-precheck",
+      description: "Skip the local STOPPED/FAILED status precheck",
+      type: "boolean",
+    },
+  ],
+  exampleArgs: ["--deployed-model dep-...", "--deployed-model dep-... --yes"],
+  async run(config: Config, flags: GlobalFlags) {
+    const deployedModel = flags.deployedModel as string | undefined;
+    if (!deployedModel) failIfMissing("deployed-model", "bl deploy delete --deployed-model <id>");
+
+    const format = detectOutputFormat(config.output);
+
+    if (config.dryRun) {
+      emitResult({ action: "deploy.delete", deployed_model: deployedModel }, format);
+      return;
+    }
+
+    // Precheck status unless skipped — surface a clear hint instead of letting
+    // the server return a generic precondition error.
+    if (!flags.skipPrecheck) {
+      try {
+        const get = await getDeployment(config, deployedModel!);
+        const deployment = get.output ?? get.data;
+        const status = (deployment?.status ?? "").toUpperCase();
+        if (status && status !== "STOPPED" && status !== "FAILED") {
+          throw new BailianError(
+            `Deployment ${deployedModel} is ${status}. Only STOPPED / FAILED deployments can be deleted. ` +
+              `Stop it first via the platform console, or pass --skip-precheck to attempt deletion anyway.`,
+            ExitCode.USAGE,
+          );
+        }
+      } catch (e) {
+        if (e instanceof BailianError) throw e;
+        // If the get itself failed (e.g. not found), let the DELETE call surface the real error.
+      }
+    }
+
+    if (!flags.yes && !config.nonInteractive && !config.quiet) {
+      process.stderr.write(`Delete deployment ${deployedModel}?\n`);
+      const ok = await promptConfirm({ message: "Proceed?", initialValue: false });
+      if (!ok) {
+        emitBare("Cancelled.");
+        return;
+      }
+    } else if (!flags.yes && config.nonInteractive) {
+      throw new BailianError(
+        "Pass --yes to confirm deletion in non-interactive mode.",
+        ExitCode.USAGE,
+      );
+    }
+
+    const response = await deleteDeployment(config, deployedModel!);
+
+    if (config.quiet) {
+      emitBare(deployedModel!);
+    } else if (format === "text") {
+      emitBare(`Deleted ${deployedModel}.`);
+    } else {
+      emitResult(response, format);
+    }
+  },
+});

packages/commands/src/commands/deploy/get.ts

@@ -0,0 +1,77 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  getDeployment,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "Get details of a single model deployment",
+  usageArgs: "--deployed-model <id>",
+  options: [
+    {
+      flag: "--deployed-model <id>",
+      description: "Deployed model identifier (required)",
+      required: true,
+    },
+  ],
+  exampleArgs: [
+    "--deployed-model qwen-plus-2025-12-01-b6d61c71",
+    "--deployed-model qwen-plus-2025-12-01-b6d61c71 --output json",
+  ],
+  async run(config: Config, flags: GlobalFlags) {
+    const deployedModel = flags.deployedModel as string | undefined;
+    if (!deployedModel) failIfMissing("deployed-model", "bl deploy get --deployed-model <id>");
+
+    const format = detectOutputFormat(config.output);
+
+    if (config.dryRun) {
+      emitResult({ action: "deploy.get", deployed_model: deployedModel }, format);
+      return;
+    }
+
+    const response = await getDeployment(config, deployedModel!);
+    const deployment = response.output ?? response.data;
+
+    if (!deployment) {
+      emitBare(`No data returned for ${deployedModel}`);
+      return;
+    }
+
+    const item: Record<string, unknown> = {
+      deployed_model: deployment.deployed_model ?? deployedModel,
+      deployed_name: deployment.name ?? "",
+      model_name: deployment.model_name ?? "",
+      base_model: deployment.base_model ?? "",
+      status: deployment.status ?? "",
+      plan: deployment.plan ?? "",
+    };
+    if (deployment.model_unit_spec) item.model_unit_spec = deployment.model_unit_spec;
+    if (deployment.charge_type) item.charge_type = deployment.charge_type;
+    if (deployment.capacity !== undefined) item.capacity = deployment.capacity;
+    if (deployment.base_capacity !== undefined) item.base_capacity = deployment.base_capacity;
+    if (deployment.ready_capacity !== undefined) item.ready_capacity = deployment.ready_capacity;
+    if (deployment.rpm_limit !== undefined) item.rpm_limit = deployment.rpm_limit;
+    if (deployment.tpm_limit !== undefined) item.tpm_limit = deployment.tpm_limit;
+    if (deployment.input_tpm !== undefined) item.input_tpm = deployment.input_tpm;
+    if (deployment.output_tpm !== undefined) item.output_tpm = deployment.output_tpm;
+    if (deployment.gmt_create) item.created_at = deployment.gmt_create;
+    if (deployment.gmt_modified) item.updated_at = deployment.gmt_modified;
+
+    if (format === "json") {
+      emitResult(item, format);
+      return;
+    }
+
+    // text / quiet — fixed-width label column for alignment
+    const label = (key: string) => `${key}:`.padEnd(18);
+    for (const [key, value] of Object.entries(item)) {
+      if (value === "" || value === undefined) continue;
+      const display = typeof value === "string" ? value : JSON.stringify(value);
+      emitBare(`${label(key)}${display}`);
+    }
+  },
+});

packages/commands/src/commands/deploy/list.ts

@@ -0,0 +1,74 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  listDeployments,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+import { formatTable } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "List model deployments",
+  usageArgs: "[--page <n>] [--page-size <n>] [--status <s>]",
+  options: [
+    { flag: "--page <n>", description: "Page number (default: 1)", type: "number" },
+    {
+      flag: "--page-size <n>",
+      description: "Results per page (default: 10, max 100)",
+      type: "number",
+    },
+    {
+      flag: "--status <s>",
+      description: "Filter by status (PENDING / RUNNING / STOPPED / FAILED)",
+    },
+  ],
+  exampleArgs: ["", "--status RUNNING", "--page-size 20 --output json"],
+  async run(config: Config, flags: GlobalFlags) {
+    const format = detectOutputFormat(config.output);
+    const pageNo = flags.page !== undefined ? (flags.page as number) : undefined;
+    const pageSize = flags.pageSize !== undefined ? (flags.pageSize as number) : undefined;
+    const status = (flags.status as string | undefined) || undefined;
+
+    if (config.dryRun) {
+      emitResult({ action: "deploy.list", page: pageNo, page_size: pageSize, status }, format);
+      return;
+    }
+
+    const response = await listDeployments(config, { pageNo, pageSize, status });
+    const payload = response.output ?? response.data;
+    const deployments = payload?.deployments ?? [];
+    const total = payload?.total;
+
+    const items = deployments.map((item) => ({
+      deployed_model: item.deployed_model ?? "",
+      model_name: item.model_name ?? "",
+      status: item.status ?? "",
+      plan: item.plan ?? "",
+      capacity: item.capacity !== undefined ? String(item.capacity) : "",
+      created_at: item.gmt_create ?? "",
+    }));
+
+    if (format === "json") {
+      emitResult({ items, total }, format);
+      return;
+    }
+
+    // text / quiet
+    if (items.length === 0) {
+      emitBare("No deployments found.");
+      return;
+    }
+    const headers = ["DEPLOYED_MODEL", "MODEL_NAME", "STATUS", "PLAN", "CAPACITY", "CREATED_AT"];
+    const rows = items.map((i) => [
+      i.deployed_model,
+      i.model_name,
+      i.status,
+      i.plan,
+      i.capacity,
+      i.created_at,
+    ]);
+    for (const line of formatTable(headers, rows)) emitBare(line);
+    if (total !== undefined) emitBare(`\nTotal: ${total}`);
+  },
+});

packages/commands/src/commands/deploy/models.ts

@@ -0,0 +1,165 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  listDeployableModels,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+import { formatTable } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "List models available for deployment",
+  usageArgs: "[--page <n>] [--page-size <n>] [--version <v>] [--source <custom|public>]",
+  options: [
+    { flag: "--page <n>", description: "Page number (default: 1)", type: "number" },
+    {
+      flag: "--page-size <n>",
+      description: "Results per page (default: 100)",
+      type: "number",
+    },
+    {
+      flag: "--version <v>",
+      description: "Catalog version filter (default: v1.0; required for new catalog models)",
+    },
+    {
+      flag: "--source <s>",
+      description: "Model source filter: custom (fine-tuned) | base (catalog) | public",
+    },
+  ],
+  exampleArgs: [
+    "",
+    "--source base",
+    "--source custom --page-size 50",
+    "--version v1.0 --output json",
+  ],
+  async run(config: Config, flags: GlobalFlags) {
+    const format = detectOutputFormat(config.output);
+    const pageNo = flags.page !== undefined ? (flags.page as number) : undefined;
+    const pageSize = flags.pageSize !== undefined ? (flags.pageSize as number) : undefined;
+    // Default version to v1.0 — without it, the API returns the legacy catalog
+    // (only old fine-tune outputs). Pass --version "" to opt out.
+    const version =
+      flags.version === "" ? undefined : ((flags.version as string | undefined) ?? "v1.0");
+    const modelSource = (flags.source as string | undefined) || undefined;
+
+    if (config.dryRun) {
+      emitResult(
+        {
+          action: "deploy.models",
+          page: pageNo,
+          page_size: pageSize,
+          version,
+          model_source: modelSource,
+        },
+        format,
+      );
+      return;
+    }
+
+    const response = await listDeployableModels(config, {
+      pageNo,
+      pageSize,
+      version,
+      modelSource,
+    });
+    const payload = response.output ?? response.data;
+    const models = payload?.models ?? [];
+    const total = payload?.total;
+
+    // Two response shapes:
+    //   - custom (fine-tuned): top-level supported_plans: string[]
+    //   - base (catalog):      plans: [{plan, templates?, cu_specs?}]
+    // For json: surface the deployment-relevant fields preserved as a tree, so
+    // downstream tooling can drive `bl deploy create --template-id <…>` without
+    // a second round-trip. For text: keep the compact one-line summary.
+    if (format === "json") {
+      const items = models.map((m) => {
+        const out: Record<string, unknown> = {
+          model_name: m.model_name ?? "",
+        };
+        if (m.base_model) out.base_model = m.base_model;
+        if (m.model_source) out.model_source = m.model_source;
+        if (m.supported_plans && m.supported_plans.length > 0) {
+          out.supported_plans = m.supported_plans;
+        }
+        if (m.plans && m.plans.length > 0) {
+          out.plans = m.plans.map((p) => {
+            const planEntry: Record<string, unknown> = { plan: p.plan ?? "" };
+            if (p.cu_specs && p.cu_specs.length > 0) {
+              planEntry.cu_specs = p.cu_specs;
+            }
+            if (p.templates && p.templates.length > 0) {
+              // Pull the top 6 fields most useful for `bl deploy create`.
+              // Drop noisy/redundant: template_source, template_type,
+              // template_version, deploy_spec (typically == template_id).
+              planEntry.templates = p.templates.map((t) => {
+                const tpl: Record<string, unknown> = {};
+                if (t.template_id) tpl.template_id = t.template_id;
+                if (t.template_name) tpl.template_name = t.template_name;
+                if (t.charge_type) tpl.charge_type = t.charge_type;
+                // Flatten roles.unified for the common COUPLED case.
+                const unified = t.roles?.unified;
+                if (unified?.model_unit_spec) tpl.model_unit_spec = unified.model_unit_spec;
+                if (unified?.capacity_unit_per_instance !== undefined)
+                  tpl.capacity_unit_per_instance = unified.capacity_unit_per_instance;
+                // Preserve split-role configs (SEPERATED) as-is so callers
+                // can still drive prefill/decode sizing.
+                if (t.roles?.prefill || t.roles?.decode) {
+                  tpl.roles = {
+                    prefill: t.roles?.prefill,
+                    decode: t.roles?.decode,
+                  };
+                }
+                if (t.template_desc) tpl.template_desc = t.template_desc;
+                return tpl;
+              });
+            }
+            return planEntry;
+          });
+        }
+        return out;
+      });
+      emitResult({ items, total }, format);
+      return;
+    }
+
+    // text / quiet — keep the compact single-line summary table.
+    const textItems = models.map((m) => {
+      let plansSummary = "";
+      if (m.supported_plans && m.supported_plans.length > 0) {
+        plansSummary = m.supported_plans.join(",");
+      } else if (m.plans && m.plans.length > 0) {
+        plansSummary = m.plans
+          .map((p) => {
+            const planName = p.plan ?? "?";
+            if (p.templates && p.templates.length > 0) {
+              return `${planName}(${p.templates.length}t)`;
+            }
+            if (p.cu_specs && p.cu_specs.length > 0) {
+              return `${planName}(${p.cu_specs.join("/")})`;
+            }
+            return planName;
+          })
+          .join(",");
+      } else {
+        plansSummary = "-";
+      }
+      return {
+        model_name: m.model_name ?? "",
+        base_model: m.base_model ?? "",
+        source: m.model_source ?? "",
+        plans: plansSummary,
+      };
+    });
+
+    if (textItems.length === 0) {
+      emitBare("No deployable models found.");
+      return;
+    }
+    const headers = ["MODEL_NAME", "BASE_MODEL", "SOURCE", "PLANS"];
+    const rows = textItems.map((i) => [i.model_name, i.base_model, i.source, i.plans]);
+    for (const line of formatTable(headers, rows)) emitBare(line);
+    if (total !== undefined) emitBare(`\nTotal: ${total}`);
+  },
+});

packages/commands/src/commands/deploy/plans.ts

@@ -0,0 +1,230 @@
+/**
+ * Per-plan strategy table for `bl deploy create`.
+ *
+ * Each PlanStrategy owns one slice of plan-specific behaviour:
+ *   - required-flag checks (USAGE errors when the user is missing something)
+ *   - any pre-flight side-effects (e.g. mu auto-picks a template from the
+ *     catalog; lora/ptu are pure)
+ *   - the plan-specific body fragment for POST /api/v1/deployments
+ *   - the plan-specific confirmation-panel rows
+ *
+ * The dispatcher in `create.ts` only knows about `STRATEGIES[plan]`. Adding a
+ * new plan = one new strategy object + one line in `STRATEGIES`. Nothing in
+ * `create.ts` needs to change. This collapses the 5 places where lora / ptu /
+ * mu used to be hard-coded (default value list / required-flag checks /
+ * auto-pick / body assembly / confirm rows) into one strategy entry per plan.
+ */
+import {
+  listDeployableModels,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing } from "bailian-cli-runtime";
+
+export interface PlanContext {
+  config: Config;
+  flags: GlobalFlags;
+  /** Underlying model identifier (`--model`). */
+  model: string;
+  /** Console display name (`--name`). */
+  name: string;
+}
+
+export interface PlanResolved {
+  /**
+   * Plan-specific fields to merge into the request body. The shared envelope
+   * (`{model_name, name, plan}`) is added by the caller.
+   */
+  body: Record<string, unknown>;
+  /**
+   * Lines to append to the confirmation panel — each already formatted like
+   * `  key:        value`.
+   */
+  confirmRows: string[];
+  /**
+   * Suffix appended to the `plan: <name>` confirm row, e.g.
+   * ` (Token-billed)`. Empty / undefined when no annotation is needed.
+   */
+  planLabelSuffix?: string;
+}
+
+export interface PlanStrategy {
+  /** Plan id, matches `--plan` CLI value. */
+  name: string;
+  /** Throws USAGE-coded BailianError when required flags are missing. */
+  validateFlags(flags: GlobalFlags): void;
+  /**
+   * Resolve plan-specific bits to a body fragment + confirm rows. May call
+   * into the API (e.g. mu auto-picks a template from the deployable-models
+   * catalog).
+   */
+  resolve(ctx: PlanContext): Promise<PlanResolved>;
+}
+
+/**
+ * `lora` (Token-billed) — the CLI default. The API requires `capacity` even
+ * though it is ignored for token-billed plans (per the working example), so
+ * the CLI injects `1` as a placeholder.
+ */
+const loraStrategy: PlanStrategy = {
+  name: "lora",
+  validateFlags() {
+    /* no required flags */
+  },
+  async resolve(): Promise<PlanResolved> {
+    return {
+      body: { capacity: 1 },
+      confirmRows: [],
+      planLabelSuffix: " (Token-billed)",
+    };
+  },
+};
+
+/**
+ * `ptu` (Token-billed, provisioned throughput). The platform rejects creation
+ * without `ptu_capacity.input_tpm` / `output_tpm` ("Miss ptu capacity info")
+ * even though the doc lists 10000/1000 defaults — so the CLI treats them as
+ * required.
+ */
+const ptuStrategy: PlanStrategy = {
+  name: "ptu",
+  validateFlags(flags: GlobalFlags): void {
+    const usage =
+      "bl deploy create --plan ptu --model <m> --name <n> --input-tpm <n> --output-tpm <n>";
+    if (flags.inputTpm === undefined) failIfMissing("input-tpm", usage);
+    if (flags.outputTpm === undefined) failIfMissing("output-tpm", usage);
+  },
+  async resolve(ctx: PlanContext): Promise<PlanResolved> {
+    const inputTpm = ctx.flags.inputTpm as number;
+    const outputTpm = ctx.flags.outputTpm as number;
+    const thinkingOutputTpm = ctx.flags.thinkingOutputTpm as number | undefined;
+    const ptuCapacity: Record<string, number> = {
+      input_tpm: inputTpm,
+      output_tpm: outputTpm,
+    };
+    if (thinkingOutputTpm !== undefined) ptuCapacity.thinking_output_tpm = thinkingOutputTpm;
+
+    const rows = [`  input_tpm:    ${inputTpm}`, `  output_tpm:   ${outputTpm}`];
+    if (thinkingOutputTpm !== undefined) rows.push(`  thinking_output_tpm: ${thinkingOutputTpm}`);
+
+    return {
+      body: { ptu_capacity: ptuCapacity },
+      confirmRows: rows,
+      planLabelSuffix: " (Token-billed, provisioned throughput)",
+    };
+  },
+};
+
+/**
+ * `mu` (model-unit-billed). `capacity`, `billing_method` and `template_id` are
+ * all required by the API but every one has a CLI-side default:
+ *   - billing_method defaults to POST_PAY (the only supported value).
+ *   - template_id auto-picks from GET /deployments/models — the one whose
+ *     `charge_type` matches `billing_method`, else the first available.
+ *   - capacity defaults to the template's `capacity_unit_per_instance` (the
+ *     smallest valid multiple of base_capacity).
+ *
+ * The catalog lookup is skipped when `--template-id` is supplied explicitly:
+ * fine-tuned custom models may not appear in the `source=base` catalog, and
+ * forcing the lookup would otherwise raise a spurious "no template" error.
+ * It is also skipped in dry-run mode to keep `--dry-run` side-effect-free.
+ */
+const muStrategy: PlanStrategy = {
+  name: "mu",
+  validateFlags() {
+    /* every required field has a default — nothing to assert up-front */
+  },
+  async resolve(ctx: PlanContext): Promise<PlanResolved> {
+    const billingMethod = (ctx.flags.billingMethod as string | undefined) || "POST_PAY";
+    let templateId = ctx.flags.templateId as string | undefined;
+    let capacity = ctx.flags.capacity as number | undefined;
+    let autoPickedTemplate = false;
+
+    if (!ctx.config.dryRun && !templateId) {
+      try {
+        const resp = await listDeployableModels(ctx.config, {
+          modelSource: "base",
+          pageSize: 100,
+          version: "v1.0",
+        });
+        const payload = resp.output ?? resp.data;
+        const target = (payload?.models ?? []).find((m) => m.model_name === ctx.model);
+        const muPlan = target?.plans?.find((p) => p.plan === "mu");
+        const templates = muPlan?.templates ?? [];
+        if (templates.length === 0) {
+          throw new BailianError(
+            `No mu-plan template found for model "${ctx.model}". ` +
+              `Run \`bl deploy models --source base\` to inspect available models, ` +
+              `or pass --template-id explicitly.`,
+            ExitCode.USAGE,
+          );
+        }
+        // POST_PAY → post_paid template; fall back to the first available.
+        const wantChargeType = billingMethod === "POST_PAY" ? "post_paid" : "pre_paid";
+        const picked = templates.find((t) => t.charge_type === wantChargeType) ?? templates[0];
+        if (!picked?.template_id) {
+          throw new BailianError(
+            `No mu-plan template found for model "${ctx.model}". ` +
+              `Run \`bl deploy models --source base\` to inspect available models, ` +
+              `or pass --template-id explicitly.`,
+            ExitCode.USAGE,
+          );
+        }
+        templateId = picked.template_id;
+        autoPickedTemplate = true;
+        if (capacity === undefined) {
+          capacity = picked.roles?.unified?.capacity_unit_per_instance ?? 1;
+        }
+      } catch (e) {
+        if (e instanceof BailianError) throw e;
+        throw new BailianError(
+          `Failed to auto-pick template for plan=mu: ${(e as Error).message}. ` +
+            `Pass --template-id explicitly.`,
+          ExitCode.USAGE,
+        );
+      }
+    }
+
+    const body: Record<string, unknown> = {
+      capacity: capacity ?? 1,
+      billing_method: billingMethod,
+    };
+    if (templateId) body.template_id = templateId;
+
+    const rows: string[] = [];
+    if (templateId) {
+      const hint = autoPickedTemplate ? " (auto-picked)" : "";
+      rows.push(`  template_id:  ${templateId}${hint}`);
+    }
+    rows.push(`  capacity:     ${capacity ?? 1}`);
+    rows.push(`  billing_method: ${billingMethod}`);
+
+    return { body, confirmRows: rows };
+  },
+};
+
+/**
+ * Registry of supported plans. Adding a new plan = one entry here. The
+ * catalog lists some additional plan names (e.g. `ptu_v2`) that are NOT
+ * accepted by the create endpoint, so the dispatcher in `create.ts` will
+ * reject anything outside this table with a clear USAGE error.
+ */
+export const STRATEGIES: Record<string, PlanStrategy> = {
+  lora: loraStrategy,
+  ptu: ptuStrategy,
+  mu: muStrategy,
+};
+
+/** Throws USAGE if `plan` is not in the strategy table. */
+export function pickPlanStrategy(plan: string): PlanStrategy {
+  const s = STRATEGIES[plan];
+  if (!s) {
+    throw new BailianError(
+      `Unsupported plan "${plan}". Supported plans: ${Object.keys(STRATEGIES).join(", ")}.`,
+      ExitCode.USAGE,
+    );
+  }
+  return s;
+}

packages/commands/src/commands/deploy/scale.ts

@@ -0,0 +1,106 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  scaleDeployment,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing, promptConfirm } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+/**
+ * `bl deploy scale` — adjust capacity (and optional PTU input/output token rates).
+ *
+ * Server-side capacity constraint: positive integer, < 1000, must be an
+ * integer multiple of `base_capacity` (visible via `bl deploy get`).
+ */
+export default defineCommand({
+  description: "Scale a deployment's capacity",
+  usageArgs: "--deployed-model <id> --capacity <n> [--input-tpm <n>] [--output-tpm <n>] [--yes]",
+  options: [
+    {
+      flag: "--deployed-model <id>",
+      description: "Deployed model identifier (required)",
+      required: true,
+    },
+    {
+      flag: "--capacity <n>",
+      description: "New capacity in plan units (must be a multiple of base_capacity)",
+      type: "number",
+    },
+    {
+      flag: "--input-tpm <n>",
+      description: "PTU only — input tokens per minute",
+      type: "number",
+    },
+    {
+      flag: "--output-tpm <n>",
+      description: "PTU only — output tokens per minute",
+      type: "number",
+    },
+    { flag: "--yes", description: "Skip the confirmation prompt", type: "boolean" },
+  ],
+  exampleArgs: [
+    "--deployed-model qwen-plus-...-b6d61c71 --capacity 8",
+    "--deployed-model dep-... --capacity 2 --yes",
+  ],
+  async run(config: Config, flags: GlobalFlags) {
+    const deployedModel = flags.deployedModel as string | undefined;
+    if (!deployedModel)
+      failIfMissing("deployed-model", "bl deploy scale --deployed-model <id> --capacity <n>");
+
+    const capacity = flags.capacity !== undefined ? (flags.capacity as number) : undefined;
+    const inputTpm = flags.inputTpm !== undefined ? (flags.inputTpm as number) : undefined;
+    const outputTpm = flags.outputTpm !== undefined ? (flags.outputTpm as number) : undefined;
+
+    if (capacity === undefined && inputTpm === undefined && outputTpm === undefined) {
+      throw new BailianError(
+        "Provide at least one of --capacity / --input-tpm / --output-tpm.",
+        ExitCode.USAGE,
+      );
+    }
+
+    const format = detectOutputFormat(config.output);
+    const body: Record<string, unknown> = {};
+    if (capacity !== undefined) body.capacity = capacity;
+    if (inputTpm !== undefined) body.input_tpm = inputTpm;
+    if (outputTpm !== undefined) body.output_tpm = outputTpm;
+
+    if (config.dryRun) {
+      emitResult({ action: "deploy.scale", deployed_model: deployedModel, body }, format);
+      return;
+    }
+
+    if (!flags.yes && !config.nonInteractive && !config.quiet) {
+      const parts: string[] = [];
+      if (capacity !== undefined) parts.push(`capacity=${capacity}`);
+      if (inputTpm !== undefined) parts.push(`input_tpm=${inputTpm}`);
+      if (outputTpm !== undefined) parts.push(`output_tpm=${outputTpm}`);
+      process.stderr.write(`Scale deployment ${deployedModel} (${parts.join(", ")})?\n`);
+      const ok = await promptConfirm({ message: "Proceed?", initialValue: false });
+      if (!ok) {
+        emitBare("Cancelled.");
+        return;
+      }
+    } else if (!flags.yes && config.nonInteractive) {
+      throw new BailianError(
+        "Pass --yes to confirm scaling in non-interactive mode.",
+        ExitCode.USAGE,
+      );
+    }
+
+    const response = await scaleDeployment(config, deployedModel!, body);
+    const deployment = response.output ?? response.data;
+
+    if (config.quiet) {
+      emitBare(deployedModel!);
+    } else if (format === "text") {
+      const cap = deployment?.capacity !== undefined ? ` (capacity=${deployment.capacity})` : "";
+      emitBare(`Scaled ${deployedModel}${cap}.`);
+    } else {
+      emitResult(response, format);
+    }
+  },
+});

packages/commands/src/commands/deploy/update.ts

@@ -0,0 +1,99 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  updateDeployment,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing, promptConfirm } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+/**
+ * `bl deploy update` — update deployment rate limits.
+ *
+ * PUT /api/v1/deployments/{deployed_model}
+ * Body: at least one of `rpm_limit` (requests/min) or `tpm_limit` (tokens/min).
+ */
+export default defineCommand({
+  description: "Update a deployment's rate limits (rpm_limit / tpm_limit)",
+  usageArgs: "--deployed-model <id> [--rpm-limit <n>] [--tpm-limit <n>] [--yes]",
+  options: [
+    {
+      flag: "--deployed-model <id>",
+      description: "Deployed model identifier (required)",
+      required: true,
+    },
+    {
+      flag: "--rpm-limit <n>",
+      description: "Requests per minute",
+      type: "number",
+    },
+    {
+      flag: "--tpm-limit <n>",
+      description: "Tokens per minute",
+      type: "number",
+    },
+    { flag: "--yes", description: "Skip the confirmation prompt", type: "boolean" },
+  ],
+  exampleArgs: [
+    "--deployed-model dep-... --rpm-limit 1000",
+    "--deployed-model dep-... --rpm-limit 1000 --tpm-limit 200000 --yes",
+  ],
+  notes: ["At least one of --rpm-limit / --tpm-limit must be provided."],
+  async run(config: Config, flags: GlobalFlags) {
+    const deployedModel = flags.deployedModel as string | undefined;
+    if (!deployedModel)
+      failIfMissing("deployed-model", "--deployed-model <id> [--rpm-limit <n>] [--tpm-limit <n>]");
+
+    const rpmLimit = flags.rpmLimit !== undefined ? (flags.rpmLimit as number) : undefined;
+    const tpmLimit = flags.tpmLimit !== undefined ? (flags.tpmLimit as number) : undefined;
+
+    if (rpmLimit === undefined && tpmLimit === undefined) {
+      throw new BailianError("Provide at least one of --rpm-limit / --tpm-limit.", ExitCode.USAGE);
+    }
+
+    const format = detectOutputFormat(config.output);
+    const body: Record<string, unknown> = {};
+    if (rpmLimit !== undefined) body.rpm_limit = rpmLimit;
+    if (tpmLimit !== undefined) body.tpm_limit = tpmLimit;
+
+    if (config.dryRun) {
+      emitResult({ action: "deploy.update", deployed_model: deployedModel, body }, format);
+      return;
+    }
+
+    if (!flags.yes && !config.nonInteractive && !config.quiet) {
+      const parts: string[] = [];
+      if (rpmLimit !== undefined) parts.push(`rpm_limit=${rpmLimit}`);
+      if (tpmLimit !== undefined) parts.push(`tpm_limit=${tpmLimit}`);
+      process.stderr.write(`Update rate limits for ${deployedModel} (${parts.join(", ")})?\n`);
+      const ok = await promptConfirm({ message: "Proceed?", initialValue: false });
+      if (!ok) {
+        emitBare("Cancelled.");
+        return;
+      }
+    } else if (!flags.yes && config.nonInteractive) {
+      throw new BailianError(
+        "Pass --yes to confirm rate-limit update in non-interactive mode.",
+        ExitCode.USAGE,
+      );
+    }
+
+    const response = await updateDeployment(config, deployedModel!, body);
+    const deployment = response.output ?? response.data;
+
+    if (config.quiet) {
+      emitBare(deployedModel!);
+    } else if (format === "text") {
+      const parts: string[] = [];
+      if (deployment?.rpm_limit !== undefined) parts.push(`rpm_limit=${deployment.rpm_limit}`);
+      if (deployment?.tpm_limit !== undefined) parts.push(`tpm_limit=${deployment.tpm_limit}`);
+      const summary = parts.length ? ` (${parts.join(", ")})` : "";
+      emitBare(`Updated ${deployedModel}${summary}.`);
+    } else {
+      emitResult(response, format);
+    }
+  },
+});

packages/commands/src/commands/file/upload.ts

@@ -6,14 +6,12 @@ import {
   type GlobalFlags,
   uploadFile,
 } from "bailian-cli-core";
-import { failIfMissing } from "../../output/prompt.ts";
-import { emitResult, emitBare } from "../../output/output.ts";
+import { failIfMissing, cmdUsage } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
 
 export default defineCommand({
-  name: "file upload",
   description: "Upload a local file to DashScope temporary storage (48h)",
-  apiDocs: "/developer-reference/get-temporary-file-url",
-  usage: "bl file upload --file <path> --model <model>",
+  usageArgs: "--file <path> --model <model>",
   options: [
     {
       flag: "--file <path>",
@@ -26,21 +24,21 @@ export default defineCommand({
       required: true,
     },
   ],
-  examples: [
-    "bl file upload --file photo.jpg --model qwen3-vl-plus",
-    "bl file upload --file video.mp4 --model wan2.1-t2v-plus",
-    "bl file upload --file audio.wav --model qwen3-asr-flash",
-    "bl file upload --file cat.png --model qwen-image-2.0",
+  exampleArgs: [
+    "--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",
   ],
   async run(config: Config, flags: GlobalFlags) {
     const filePath = flags.file as string | undefined;
     if (!filePath) {
-      failIfMissing("file", "bl file upload --file <path> --model <model>");
+      failIfMissing("file", cmdUsage(config, "--file <path> --model <model>"));
     }
 
     const model = flags.model as string | undefined;
     if (!model) {
-      failIfMissing("model", "bl file upload --file <path> --model <model>");
+      failIfMissing("model", cmdUsage(config, "--file <path> --model <model>"));
     }
 
     const format = detectOutputFormat(config.output);

packages/commands/src/commands/finetune/cancel.ts

@@ -0,0 +1,62 @@
+import {
+  defineCommand,
+  detectOutputFormat,
+  cancelFineTune,
+  BailianError,
+  ExitCode,
+  type Config,
+  type GlobalFlags,
+} from "bailian-cli-core";
+import { failIfMissing, promptConfirm } from "bailian-cli-runtime";
+import { emitResult, emitBare } from "bailian-cli-runtime";
+
+export default defineCommand({
+  description: "Cancel a running fine-tune job",
+  usageArgs: "--job-id <id> [--yes]",
+  options: [
+    { flag: "--job-id <id>", description: "Fine-tune job ID (required)", required: true },
+    { flag: "--yes", description: "Skip the confirmation prompt", type: "boolean" },
+  ],
+  exampleArgs: ["bl finetune cancel --job-id ft-xxx", "bl finetune cancel --job-id ft-xxx --yes"],
+  notes: [
+    "Only PENDING / RUNNING jobs can be cancelled. Completed / failed / already-",
+    "cancelled jobs return a server-side error (passed through verbatim).",
+  ],
+  async run(config: Config, flags: GlobalFlags) {
+    const jobId = flags.jobId as string | undefined;
+    if (!jobId) failIfMissing("job-id", "bl finetune cancel --job-id <id>");
+
+    const format = detectOutputFormat(config.output);
+
+    if (config.dryRun) {
+      emitResult({ action: "finetune.cancel", job_id: jobId }, format);
+      return;
+    }
+
+    if (!flags.yes && !config.nonInteractive && !config.quiet) {
+      process.stderr.write(`Cancel fine-tune job ${jobId}?\n`);
+      const ok = await promptConfirm({ message: "Proceed?", initialValue: false });
+      if (!ok) {
+        emitBare("Cancelled.");
+        return;
+      }
+    } else if (!flags.yes && config.nonInteractive) {
+      throw new BailianError(
+        "Pass --yes to confirm cancellation in non-interactive mode.",
+        ExitCode.USAGE,
+      );
+    }
+
+    const response = await cancelFineTune(config, jobId!);
+    const job = response.output ?? response.data;
+
+    if (config.quiet) {
+      emitBare(jobId!);
+    } else if (format === "text") {
+      const status = job?.status ? ` (status=${job.status})` : "";
+      emitBare(`Cancelled ${jobId}${status}.`);
+    } else {
+      emitResult(response, format);
+    }
+  },
+});