Structured context for AI-powered development
AI coding assistants are incredible, but they have a fundamental limitation: they don't know your project. When an AI looks at your codebase, it sees:
import { useState } from 'react'
import axios from 'axios'But it doesn't know:
- Which React version you're using and its specific quirks
- How you actually use axios across your project
- Your team's established patterns and conventions
- Which parts of a library are safe to use vs. deprecated
The result? Hallucinated APIs, outdated suggestions, and wasted tokens as AI tools try to reverse-engineer your project's intent from thousands of lines of code.
README.ai is a machine-readable manifest that lives alongside your code. It provides AI tools with structured, curated context about:
- 📦 Dependencies - Not just versions, but how you use them
- 🏗️ Architecture - High-level structure and patterns
- 🔧 Code Conventions - Your team's established practices
- 🔄 Cross-project References - How microservices and libraries relate
# One command to set it up
npx readme-ai init
# Automatic updates with every commit
git commit -m "Add new API endpoint"
# README.ai updates itself! ✨When an AI encounters your project:
- It detects
README.aiautomatically - Parses the structured context
- Understands your project in seconds, not minutes
- Gives you better, more accurate assistance
- AI: "I see you're using axios. Here's how to make a GET request..."
+ AI: "I see you're using axios 1.6.0 with your custom interceptor pattern from apiClient.js. Here's how to add that new endpoint following your established error handling..."README.ai aims to become the universal standard for AI-context in software projects. We're building:
- 🔄 Auto-generation tools that keep context in sync with code
- 🔌 Universal discovery so any AI tool can find and use README.ai
- 📚 Ecosystem of schemas for popular libraries and frameworks
- 🤝 Integration with existing AI coding assistants
We're in the very early stages! This is a placeholder repository while we scope out the project. We're exploring:
- The optimal schema format (JSON? YAML? TOML?)
- Git hooks for automatic updates
- Integration with popular AI tools (Continue, Cursor, etc.)
- Community-driven library profiles
- Phase 0: Discovery - Define the schema and tooling approach
- Phase 1: Foundation - Basic auto-generation from diffs
- Phase 2: Integration - Partner with AI tool developers
- Phase 3: Ecosystem - Community library profiles and standards
We're looking for:
- Early adopters to try prototypes and provide feedback
- AI tool developers interested in integration
- Schema designers to help define the format
- Library maintainers to create official README.ai profiles
- ⭐ Star this repo to show interest
- 🗣️ Join the discussion in Issues
- 💡 Share your ideas for what README.ai should include
- 🔧 Help us build the tooling
MIT © README.ai Contributors