Muninn
The universal memory layer for your AI agent stack.
5 followers
The universal memory layer for your AI agent stack.
5 followers
Muninn gives your AI agents a brain. It indexes your projects into local Markdown files, enabling surgically precise, token-efficient context retrieval across all your tools (Claude, Cursor, etc.). Local-first, Rust-powered, and 95% token savings.
Free
Launch Team / Built With
Why we built Muninn – Bringing Odin's ravens to your terminal.
Hi Product Hunt! I'm Blackknight, and I'm tired of my AI agents forgetting what we worked on yesterday.
Current AI agents have a "goldfish memory" problem. To give them context, you often have to dump entire files into the prompt, which is slow, expensive, and messy.
We built Muninn to solve this. Inspired by Odin's raven "Memory", it flies through your filesystem, indexes your knowledge into transparent Markdown files, and surgically injects only the most relevant facts into your AI's context window.
Key Features:
🚀 Universal: One memory for Claude, Cursor, and any MCP-compatible tool.
💸 Token Efficient: Save up to 95% on context tokens.
🔒 Local-First: Your data stays in ~/.muninn. It's just Markdown.
⚡ High Performance: Powered by the CXP Rust engine for instant retrieval.
I’d love to hear your feedback! What’s the one thing you wish your AI agent remembered better?
— Blackknight
Ps:In future is a extension planned that bundels everything and makes it visuell and easier
Am right now making an Mcp to automate everything hope you will like it
Here's my new account https://x.com/MunninDev