AI Assistant Integration¶
RoomKit provides documentation optimized for AI coding assistants, helping developers get accurate, working code when using tools like Claude Code, Cursor, GitHub Copilot, and others.
llms.txt¶
The /llms.txt file provides a structured index of RoomKit documentation following the llms.txt specification.
Location: https://your-docs-site/llms.txt
AI assistants can use this file to:
- Quickly find relevant documentation
- Understand the project structure
- Navigate to specific API references
Format¶
# RoomKit
> RoomKit is a pure async Python library for building multi-channel conversation systems...
## Getting Started
- [Features Overview](docs/features.md): Core features, channels, hooks...
## Core API
- [RoomKit](docs/api/roomkit.md): Central orchestrator class...
AGENTS.md¶
The AGENTS.md file in the repository root provides project-specific guidance for AI coding assistants.
Location: Repository root (/AGENTS.md)
What It Contains¶
| Section | Purpose |
|---|---|
| Quick Reference | Dev commands (uv sync, pytest, ruff) |
| Project Structure | Directory tree with descriptions |
| Architecture Patterns | ABC + Default, Channel, Provider, Hook patterns |
| Code Style | Required patterns, type hints, imports |
| Testing Patterns | Test class structure, fixtures |
| Common Tasks | Adding providers, channels, hooks |
| Boundaries | Always do / Ask first / Never do rules |
| Key Concepts | Room lifecycle, event flow, identity resolution |
Why It Matters¶
Without AGENTS.md:
- AI assistants guess patterns from training data
- Code may not follow RoomKit's async-first architecture
- Missing exports, wrong patterns, inconsistent style
With AGENTS.md:
- AI follows RoomKit's actual patterns
- Correct ABC + default implementation pattern
- Proper hook usage (BEFORE_BROADCAST vs AFTER_BROADCAST)
- Knows to export public classes from
__init__.py
Usage with AI Assistants¶
Claude Code / Cursor¶
These tools automatically read AGENTS.md from the repository root when working on your project.
Explicit Reference¶
You can also explicitly instruct AI assistants:
MCP Server (Future)¶
A Model Context Protocol server for live documentation search is planned. This will allow AI assistants to:
- Search RoomKit documentation in real-time
- Get up-to-date API signatures
- Find code examples across the repository
Best Practices for Contributors¶
When contributing to RoomKit:
- Follow AGENTS.md patterns - AI assistants will expect these patterns
- Update documentation - Keep llms.txt and API docs in sync
- Export public classes - Add to
roomkit/__init__.py - Add tests - Follow the testing patterns in AGENTS.md