Go AI-Friendly with Mimir
Mimir is a contextual RAG system that transforms your codebase and documentation into an intelligent knowledge base. Connect your AI assistants to your entire codebase and docs with full context understanding.
Production Ready
MCP Integration
Fast & Scalable
Open Source
Why Teams Choose Mimir
Contextual Understanding
Unlike basic RAG, Mimir provides rich context around each code entity—full file content, imports, parent classes, and surrounding code. Your AI assistant understands not just what code does, but how it fits into your system.
AI Assistant Integration
Connect Claude Code, VS Code, Claude Desktop, and other MCP-compatible assistants directly to your codebase. No more copy-pasting code—your AI assistant has instant access to your entire knowledge base.
Automatic Entity Extraction
Automatically extracts and indexes functions, classes, interfaces, and more from TypeScript, Python, and other languages. Each entity includes JSDoc comments, parameters, return types, and direct links to source code.
Everything You Need to Go AI-Friendly
Multiple Repositories
Ingest from multiple code and documentation repositories. Perfect for monorepos, microservices, or teams with separate code and docs repos.
Documentation & Code
Index both MDX documentation and source code. Your AI assistant can answer questions about APIs, implementation details, and usage examples.
GitHub Webhooks
Automatic ingestion on code changes. Keep your knowledge base up-to-date without manual triggers. Set up once and forget about it.
OpenAI-Compatible API
RESTful API that works with OpenAI-compatible SDKs and clients. Build custom interfaces or integrate with existing tools.
Supabase Vector Store
Built on Supabase for reliable, scalable vector storage. Fast semantic search with configurable similarity thresholds and match counts.
Multiple LLM Providers
Use OpenAI, Anthropic, Google, or Mistral for embeddings and chat. Mix providers to optimize for cost and quality.
How It Works
Ingest
Mimir fetches your code and documentation from GitHub repositories, extracts entities, and creates rich contextual embeddings.
Store
Embeddings are stored in Supabase with source URLs, metadata, and full context. Everything is indexed and ready for semantic search.
Query
Your AI assistant queries Mimir via MCP or REST API, gets relevant chunks with full context, and provides accurate answers with source links.
Ready to Go AI-Friendly?
Get your codebase and documentation queryable by AI in minutes. Start building with Mimir today.