AI Assistants

Picard.js publishes an llms.txt at

https://picard.js.org/llms.txt

It follows the llms.txt convention: a single plain-text file written for language models. It starts with a compact cheat sheet:

• which variant to use in which environment (browser, bundler, server, WebView),
• the attributes of pi-component and pi-slot,
• what each framework needs to accept the custom elements,
• the recipe for server-side rendering and the islands script, including streaming frameworks and non-JavaScript backends,
• the component lifecycle, the feed formats, and common pitfalls (CORS, CSP, --experimental-vm-modules, DOM shims in tests).

After that, it links to the relevant pages of this documentation.

Assistants have the same knowledge cutoff problem as everyone else: Picard.js is young and changes between pre-releases. Giving your assistant llms.txt is the quickest way to get code that uses the current API.

Using It

How you add it depends on your tool. Common options:

• Ask for it in the prompt, e.g., "Read https://picard.js.org/llms.txt, then integrate Picard.js into this app." Assistants that can fetch URLs (such as Claude Code) will read it first.

• Reference it in your project's instructions, e.g., in CLAUDE.md, AGENTS.md, or your editor's rules file, so every session starts with it:

  This project uses Picard.js for micro frontends.
  Before changing related code, read https://picard.js.org/llms.txt.

• Add it as documentation in editors that index docs by URL.

Pair It with a Sample

llms.txt explains the rules. The samples show them applied to a concrete framework. For the best results, also point your assistant to the sample closest to your stack, e.g., "Follow the approach of https://github.com/picardjs/samples/tree/main/sveltekit/ssr." Each sample's README explains exactly where Picard.js is integrated.