Merge branch 'main' into feat/composable-cli

Author: XXPermanentXX

.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 }}"

.node-version

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

CHANGELOG.md

@@ -6,6 +6,29 @@ The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and
 
 [中文版](CHANGELOG.zh.md) · [README](README.md) · [Contributing](CONTRIBUTING.md)
 
+## [1.4.2] - 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

CHANGELOG.zh.md

@@ -6,6 +6,29 @@
 
 [English](CHANGELOG.md) · [README](README.zh.md) · [参与贡献](CONTRIBUTING.zh.md)
 
+## [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

README.md

@@ -27,7 +27,7 @@ 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
 
@@ -38,6 +38,7 @@ Equip your AI Agent out-of-the-box with these capabilities, composable across co
 - **MCP integration** — Orchestrate Bailian MCP servers: list services, inspect tools, and invoke any tool directly from the terminal
 - **Web search** — Real-time internet retrieval for up-to-date, accurate answers
 - **Model recommendation** — Describe your scenario and get best-fit model suggestions; supports scoped search, model comparison, and alternative discovery
+- **Fine-tuning & deployment** — Upload datasets, create 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
 
@@ -54,7 +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
@@ -67,7 +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.
@@ -111,22 +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
 
+# 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 --expiring 30                           # Quotas expiring within 30 days
-bl usage free --sort remaining                        # Sort by remaining % ascending
-bl usage stats --workspace-id <id>                    # Usage overview for a workspace
-bl usage stats --model qwen-turbo --workspace-id <id> # Per-model usage
+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
-bl quota list                                         # View RPM/TPM limits for all models
-bl quota list --model qwen3.6-plus                    # View limits for a specific model
-bl quota check                                        # Current usage vs rate limits
-bl quota check --model qwen3.6-plus --period 5        # Check usage over last 5 minutes
+# 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
+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&)
@@ -156,9 +165,9 @@ Required for console capability commands (`app list`, `usage free`, `usage stats
 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.
 

README.zh.md

@@ -27,7 +27,7 @@ _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
 
@@ -38,6 +38,7 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 - **MCP 集成** — 统一调度百炼 MCP 服务：列出服务、查看工具、直接在终端调用任意工具
 - **联网搜索** — 实时互联网信息检索，提升回答准确性及时效性
 - **模型推荐** — 描述你的场景，智能推荐最适合的模型；支持限定范围搜索、模型对比和替代发现
+- **微调与部署** — 上传数据集、创建 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
 
@@ -54,7 +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
@@ -65,7 +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。
@@ -82,7 +83,10 @@ npx skills add modelstudioai/cli --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
@@ -106,22 +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
 
+# 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 --expiring 30                           # 30 天内过期的额度
-bl usage free --sort remaining                        # 按剩余百分比升序排列
-bl usage stats --workspace-id <id>                    # 指定空间的用量概览
-bl usage stats --model qwen-turbo --workspace-id <id> # 指定模型用量
+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
-bl quota list                                         # 查看所有模型的 RPM/TPM 限额
-bl quota list --model qwen3.6-plus                    # 查看指定模型限额
-bl quota check                                        # 查看当前用量 vs 限流阈值
-bl quota check --model qwen3.6-plus --period 5        # 查看最近 5 分钟用量
+# 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&)
@@ -151,9 +163,9 @@ bl text chat --api-key sk-xxxxx --message "你好"
 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。