AI Agent Guide

Use this page to quickly orient AI agents to this repository without scanning every file.

Where Agent References Live

Canonical entry files at repository root:

• llms.txt (quick reference)
• llms-full.txt (full repository guide)

Focused references are grouped under:

• llms/README.md
• llms/llms-components.txt
• llms/llms-patterns.txt
• llms/llms-core.txt
• llms/llms-providers.txt
• llms/llms-utils.txt
• llms/llms-types.txt

Plain text links above open directly as raw text files (no docs chrome).

GitHub Raw Fallback

If you want direct GitHub raw URLs:

• raw llms.txt
• raw llms-full.txt

Recommended Reading Order for Agents

1. Read llms.txt first.
2. Read one focused file for the active task area.
3. Read llms-full.txt only if broader architecture context is needed.

Task-to-File Mapping

• Component behavior and props:
  • llms/llms-components.txt
• Common usage patterns and recipes:
  • llms/llms-patterns.txt
• Runtime store/bridge internals:
  • llms/llms-core.txt
• Provider/context lifecycle:
  • llms/llms-providers.txt
• Public utility facade and helper behavior:
  • llms/llms-utils.txt
• Type contracts and typed helpers:
  • llms/llms-types.txt

Ground Rules for Agent Changes

• Preserve public API behavior unless explicitly requested.
• Keep host-aware routing semantics intact.
• Prefer docs/example improvements over internal rewrites.
• Keep current docs and versioned docs in sync when docs behavior changes.
• Validate with:
  • pnpm run typecheck
  • pnpm run test
  • pnpm run docs:check