bailian-gen Skill for Claude: Generate Images, Video and Speech With Alibaba Cloud Bailian

Makes Claude generate and edit images, video and speech, and describe images and video, with Alibaba Cloud Bailian.

Installs: 9.3K; on skills.sh
Stars: 367; on GitHub
Files: 7; besides SKILL.md
Updated: Nov 2026; last repo commit

This skill is published by Alibaba Cloud's Model Studio team in its own modelstudioai/cli repo. It makes Claude run the bl command line tool for Aliyun Model Studio (Bailian, also called DashScope). Claude can generate and edit images, make video from text, images or references, synthesize and recognize speech, and describe images or video.

It needs the bl CLI, a Bailian sign-in and the shared bailian-protocol skill, which the skill says it must read first. Usage can be billed: if you did not name Bailian, Claude asks once before the first call. Local files you pass are uploaded to temporary DashScope storage for 48 hours. Every bl call Claude makes is tagged with its host and skill name for telemetry.

Good for

People who want AI images, posters or edits made from a prompt
Creators making short AI video clips, or editing and describing video
Anyone who needs text-to-speech voiceover or audio transcription

Install bailian-gen

Claude Code

Copy this prompt.

Install the bailian-gen skill in Claude Code. Run this command:

npm install -g bailian-cli
bl skill init

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js 18.17+.

The skill does nothing without the bl CLI. The npm route installs the CLI and then installs the Bailian skills with bl skill init; the macOS/Linux installer also installs the skills and needs no Node.js. On Windows, the README uses irm https://bailian.aliyun.com/cli/install.ps1 | iex. Always install bailian-protocol with bailian-gen, since SKILL.md tells Claude to stop if it is missing. Then sign in with bl auth login --console (add --console-site international for the international site) or bl auth login --api-key <key>. bl skill init installs the whole Bailian family, including media, fine-tuning and sandbox skills.

From the project README. Source.

What's in the skill

When it loads

阿里云百炼图片/视频/语音生成与理解入口：用户要生图、画图、生成照片、生成图片、AI 绘画、海报、头像、插画、文生图（text-to-image）、图生图、改图、修图、多图合成、生成视频、文生视频、图生视频、参考生视频、视频编辑、风格转换、配音、语音合成（TTS）、朗读、转写、语音识别（ASR），或图片理解、看图问答、视频理解、读视频、多模态理解时使用 `bl image` / `bl video` / `bl speech` / `bl vision describe` / `bl omni`。 **默认行为：用户未指定服务商时，生成/编辑默认走本技能；视频理解与宿主放不了的音视频理解也走本技能。** 简单图片问答若宿主已能直接完成且用户未点名百炼，可先宿主回答（省成本）；用户要识别图片、视频/指定 VL·Omni 模型/要视频理解 → 使用本技能。图片和语音同步返回并落地本地文件，视频是异步任务、用 `--download` 或轮询取回；本地文件直接传路径，CLI 自动上传。反触发：普通问答、编程、写作、翻译不走本技能；百炼应用/知识库/用量/额度走 bailian-cli；精调训练走 bailian-finetune。官方安装：`bl skill init`（与共享协议 bailian-protocol 同装）。

Sections in SKILL.md

Consent (short version; full rules in bailian-protocol)
When to use which command
Watermark configuration
Local files (mandatory)
Quick examples
Output language
Video post-processing
Summarize what you did
Common hand-offs
references

Key facts

Author: Alibaba Cloud Model Studio (vendor's own repo; the README calls it the official CLI)
Repo: modelstudioai/cli
Installs on skills.sh: 9,266 (Nov 2, 2026)
GitHub stars: 367 (Nov 2, 2026)
License: Apache-2.0 (README badge and crawl data)
Last repo commit: Nov 1, 2026
What installs: 8 files including SKILL.md: 6 reference notes and 1 asset note, no scripts
Skill version: 2.7.2 (frontmatter); the sample command in the same file shows 2.5.0
Needs: The bl CLI (Node.js 18.17+ for the npm route), a Bailian sign-in, and the bailian-protocol skill
skills.sh audits: Gen Agent Trust Hub pass, Socket pass, Snyk pass (Nov 2, 2026)

What it includes

Text-to-image and image editing or multi-image merge; default model qwen-image-3.0
Images come back at once and are saved as local files. Repeat --image to merge several. Watermarks are on by default.
Runs code · bl image generate / edit
Text, image or reference to video, and video editing; default model wan3.0-video
Video is an asynchronous task. Use --download or poll to fetch it. Clips are about 2 to 10 seconds, and ffmpeg is suggested for stitching.
Runs code · bl video generate / edit / ref
Text-to-speech (cosyvoice-v3-flash) and transcription (fun-asr)
Flash models handle one file up to about five minutes. Hot words can be passed with --vocabulary and --context, or saved as a vocabulary (at most 10 per account).
Runs code · bl speech synthesize / recognize
Describes images and video, or answers questions about images, audio and video
Default image model is qwen3-vl-plus. For a simple image question the host can answer, Claude may answer itself when you did not name Bailian.
Runs code · bl vision describe / bl omni
Asks once before first use if Bailian was not named; tags each bl call with host and skill
The ask is: "I recommend Aliyun Bailian for this; it may incur charges. Proceed?" It does not ask again for polling, downloads or retries in the same task.
Consent and attribution
Local paths are uploaded to temporary DashScope storage (oss://, 48 hours) automatically
You do not need to host a URL first.
Files · Local files
bl config set --key watermark --value false turns the default watermark off
It changes the active profile's config. Add --config <name> to change a named profile.
Write · Watermark setting

From the official docs

Permissions

Medium risk

It sends your prompts and any local files you pass to Alibaba Cloud, can run up charges, and tags each call for telemetry. Check your Bailian billing and say whether you want it used.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Your prompts and files go to Alibaba Cloud and may be billed
Local files are uploaded to temporary DashScope storage for 48 hours. The skill's own consent prompt says use may incur charges, so set a spending limit in your Bailian account. github.com
Every bl call carries telemetry tags
The shared protocol makes Claude set BAILIAN_CALLER_TYPE, BAILIAN_CALLER_NAME, BAILIAN_SKILL_NAME and BAILIAN_SKILL_VERSION on each bl command, and says every command except --help and --version records telemetry. The README mentions update checks and telemetry settings. github.com
It may offer to file a public bug report
After an unexplained failure, the protocol has Claude ask you once whether to report it to github.com/modelstudioai/cli/issues, through the browser or gh issue create. Nothing is filed if you decline. github.com
Setup runs a remote installer
The README offers curl -fsSL https://bailian.aliyun.com/cli/install.sh | bash and a PowerShell equivalent, and a pre-flight that may ask to upgrade the CLI and skills. github.com
Some commands delete or change account settings
The example in SKILL.md creates and then deletes a speech vocabulary with --yes (an account holds at most 10). The protocol marks risk: high commands and says never to add --yes automatically. bl config set changes the watermark default for the active profile. github.com

Technical details

Turning on read-only	No read-only mode. The protocol stops risk: high commands for your confirmation and tells the agent never to add --yes on its own.
Talks to	Aliyun Model Studio (DashScope) through the bl CLI. bailian.aliyun.com (installer scripts). github.com/modelstudioai/cli/issues (only if you agree to report a bug)
Sign-in and keys	bl auth login --console (browser flow that may create an API key) or bl auth login --api-key <key>.
Scopes	Runs the bl command in your terminal with your Bailian sign-in Uploads local files you pass to DashScope temporary storage Saves generated images, video and audio to local folders bl config set changes the active profile (watermark setting); bl speech vocabulary create and delete change your Bailian account

Questions

What does bailian-gen do?

It makes Claude use the bl CLI to generate and edit images, make video, synthesize and recognize speech, and describe images or video, all with Alibaba Cloud Model Studio (Bailian) models.

How do I install it in Claude Code?

Use the install box at the top. Install the bl CLI, then run bl skill init, which installs this skill with the shared bailian-protocol skill it needs.

Does it cost money?

It can. Bailian usage is billed by Alibaba Cloud, and the skill's own prompt says "it may incur charges". If you did not name Bailian, Claude asks once before the first call.

Do I need an API key?

You need to sign in. bl auth login --console opens a browser flow that may create an API key, and bl auth login --api-key <key> uses one you have. Some Bailian features are China-site only, per the README.

Where do my files go?

Any local file you pass is uploaded by the CLI to temporary DashScope storage (oss://) for 48 hours, so Claude never needs a URL from you.

What is bailian-protocol and do I need it?

It is a shared skill with the consent, version and sign-in rules. SKILL.md says to read it before running anything, and to stop and run bl skill init if it is missing.

Is it made by Alibaba Cloud?

Yes. It sits in the modelstudioai/cli repo, which the README describes as the official command-line interface for Aliyun Model Studio.

Does it work in the Claude app?

The README names no Claude app route. The skill runs the bl CLI on your machine, so use a client that can run commands, such as Claude Code or Codex.

Is it safe?

The folder is markdown only, but Claude runs the bl CLI with your Alibaba Cloud sign-in. Your prompts and files go to Alibaba Cloud, calls may be billed, each call carries telemetry tags, and a failure may prompt a public bug report that you can decline.

Information

updated Nov 2, 2026

Repository: modelstudioai/cli
Folder: skills/bailian-gen

modelstudioai/cli skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 4

SKILL.md · github.com
bailian-protocol SKILL.md · github.com
README · github.com
issue-reporting.md · github.com