The documentation in Markdown for AI assistants (llms.txt).
AI assistants (Claude Code, GitHub Copilot, Cursor, ChatGPT...) don't need to read the pages like a person: the menu, demos and styles only use up their context. That's why the documentation is also published in Markdown, following the llms.txt format: clean text, every component's parameters and the real example code (the same these docs run), in English.
Add these lines to your assistant's instructions file (Claude Code reads CLAUDE.md, GitHub Copilot .github/copilot-instructions.md and Cursor .cursor/rules). Then, when it writes AstrolUI code, it checks the documentation instead of making up parameters. In a chat you can also paste the llms-full.txt link or the component page link.
<!-- Claude Code: CLAUDE.md · GitHub Copilot: .github/copilot-instructions.md · Cursor: .cursor/rules/astrolui.mdc -->## AstrolUIThis project uses AstrolUI (Blazor, .NET 10). Before writing UI code, read https://doc.astrolui.com/llms.txtand open the page of each component you use (https://doc.astrolui.com/ai/<page>.txt).For code written for older AstrolUI versions follow https://doc.astrolui.com/ai/migration.txtAt the bottom of every page there is a Markdown for AI link that opens the same page in AI format.
The files are generated from the pages' texts and code, so nothing is written twice. After changing the docs, regenerate them:
python 2-Shared/tools/docs_ai_export.py # llms.txt, llms-full.txt and wwwroot/ai/*.txtpython 2-Shared/tools/docs_search_index.py # search index of these docs