Astron Harness Tutorial

AStudio → Loomy → AstronClaw. Learn to turn a task into a useful deliverable, then build an AI workflow around your own work.

Read the tutorial · Start with AStudio · 简体中文

Choose your starting point

Product	Best starting point	Learn
01 · AStudio	Goal-oriented tasks and deliverables	Workspace guide · Quick start
02 · Loomy	Local knowledge and office workflows	Desktop guide · Quick start
03 · AstronClaw	Cloud tasks and messaging channels	Cloud guide · Getting started

01 · AStudio: from a goal to a deliverable

Begin with the workspace, create a task, then learn projects, Plan and Goal modes, previews, editing, permissions, automation and plugins. Follow the illustrated steps with 56 locally stored official images, shared by the Chinese and English guides.

• Core features: find the right mode and inspect the result.
• Examples and tips: practice research, website creation and iteration.
• Plans and points, FAQ and privacy: understand the rules before starting.

02 · Loomy: a companion for everyday work

Connect local files, knowledge and office tools. Explore AI companions, the toolbox, work scenarios and your personal companion.

03 · AstronClaw: keep work moving in the cloud

Start with a hosted assistant, then connect skills and messaging channels. Review scenarios and billing for the work you plan to do.

Documentation and image updates

The daily watcher checks AStudio, Loomy and AstronClaw. AStudio coverage includes all 15 official guide and agreement pages, plus the image contents. New images and replacements at the same URL are reported for review. See the watcher guide and asset sources.

Screenshots preserve the official Chinese interface; the English version supplies English captions. Images can be enlarged on the tutorial site.

Local Development

This project is built with VitePress. To preview or edit the documentation locally:

# 1. Clone the repository
git clone https://github.com/iflytek/astron-harness-tutorial.git
cd astron-harness-tutorial

# 2. Install dependencies
npm install

# 3. Start the local preview server
npm run docs:dev

# 4. Build static files
npm run docs:build

Learning Assistant model configuration

The Learning Assistant uses Spark-X2.6-4B, with streaming replies and Chinese/English pages. Copy .env.example to .env and set ASTRON_MAAS_API_KEY to a key enabled for Spark-X2.6-4B. Keep credentials server-side; do not use variables prefixed with VITE_ for API keys.

The defaults are model spark-x2.6-4b, endpoint https://maas-api.cn-huabei-1.xf-yun.com/v2/chat/completions, and low reasoning effort. Configure the server variables in .env.example to override them. Model IDs sent by cached clients cannot override the server setting.

For local chat, run npm run server in one terminal and npm run docs:dev in another. Configure the same environment variables on Vercel and redeploy. Run npm run test:chat for regression tests without external calls, or node test-api.js to verify the live model connection.

Contributing

Issues and Pull Requests are welcome! Let's polish it together and help more people bring AI into their daily workflows.

Before submitting a PR, please read our Contributing Guide. When you make your first contribution, the CLA Assistant bot will automatically leave a comment on your PR to guide you through signing the Contributor License Agreement (CLA).

Learn with the community

AStudio	Loomy	AstronClaw

AStudio Feishu group · Astron open-source community

License

Tutorial source is released under Apache License 2.0. Official product images and marks remain subject to their respective rights; see the asset notes.