Skip to content
Entangle UI

Docs for LLMs

LLMs typically have no training data about entangle-ui. To make AI-assisted development with this library reliable, we publish the documentation in plain text alongside the regular site, following the llmstxt.org convention.

Point your assistant at one of the resources below — it will get every component, hook and guide as ordinary markdown, with all live-preview JSX stripped.

Resources

  • /llms.txt — short index. One paragraph of context plus a deep-link to every page. Best when the LLM should only fetch what it needs.
  • /llms-full.txt — full documentation as a single concatenated text file. Best for piping into a context window or vector store in one shot.
  • /llms/ — per-page markdown. Useful when wiring a retriever or pulling individual references into prompts. Each MDX page in this site has a matching /llms/<slug>.md (for example, /llms/components/primitives/button.md).

The text resources are regenerated on every documentation build, so they always match what’s on this site.

Using with Claude Code

This repository ships a Claude Code skill at .claude/skills/entangle-ui/. When Claude Code runs in a project that depends on entangle-ui, the skill exposes one reference file per component, hook and guide. Drop the directory into your own .claude/skills/ (or symlink it from your node_modules/entangle-ui) and Claude will load only the file for the component it’s working with.

.claude/skills/entangle-ui/
  SKILL.md
  components/
    primitives/button.md
    primitives/input.md
    layout/flex.md
    ...
  hooks/
    use-disclosure.md
    ...
  guides/
    theming.md
    styling.md
    accessibility.md
    animations.md

Using with other assistants

Most assistants accept a URL or a pasted text block. Two recipes that work today:

  1. System prompt / project instructions — paste the contents of /llms.txt so the model knows which deep-links to fetch on demand.
  2. Single-shot context dump — paste the contents of /llms-full.txt (or attach it as a file). It is plain markdown and well under typical long-context limits.

For agents with web access, just point them at https://entangle-ui.dev/llms.txt and let them follow the links.