Mimir

Introduction

Welcome to Mimir documentation

Welcome to Mimir! Mimir is a contextual RAG system that makes your codebase and documentation queryable by AI assistants.

What is Mimir?

Mimir ingests documentation (MDX) and source code (TypeScript, Python, and more) from GitHub repositories into a Supabase vector store. Unlike basic RAG, contextual RAG provides rich context around each code entity, including full file content, imports, and surrounding code.

Key Features

  • Contextual RAG: Rich context around code entities for better understanding
  • Multiple Repository Support: Ingest from multiple code and documentation repositories
  • Language Support: TypeScript, Python, and easily extensible to more languages
  • MCP Integration: Connect AI assistants via Model Context Protocol
  • OpenAI-Compatible API: Works with OpenAI-compatible SDKs and clients
  • Automatic Entity Extraction: Extracts functions, classes, interfaces, and more

Quick Start

Get started in minutes:

  1. Set up Supabase
  2. Configure mimir-rag
  3. Start the server
  4. Ingest your first repository

See the Getting Started guide for detailed instructions.

Documentation

How It Works

┌─────────────────────┐
│   GitHub Repo       │
│   (Documentation)   │
└──────────┬──────────┘

           │ Webhook / Manual Ingest

┌─────────────────────┐
│   mimir-rag         │
│   - Ingestion       │
│   - Chunking        │
│   - Embedding       │
│   - Vector Store    │
└──────────┬──────────┘

           │ /mcp/ask endpoint

┌─────────────────────┐       ┌─────────────────────┐
│   Your MCP Server   │◄──────┤  AI Assistant       │
│   (Custom Package)  │       │  - Claude Code      │
└─────────────────────┘       │  - VSCode           │
                              │  - Claude Desktop   │
                              └─────────────────────┘

Use Cases

  • AI-Powered Code Assistant: Query your codebase in real-time
  • Documentation Search: Make your docs instantly queryable
  • Codebase Understanding: Index entire codebases with full context
  • Internal Knowledge Base: Index wikis, API docs, and technical documentation
  • Developer Onboarding: Help new developers find information quickly