Uranid

Uranid / mnem

Public

Git for Knowledge Graphs: versioned agent memory with hybrid GraphRAG retrieval. Runs entirely offline, no LLM required.

agent-memory graphrag knowledge-graph local-first mcp
17
3
100% credibility
Found May 03, 2026 at 17 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

mnem provides a Git-like system for creating and querying versioned knowledge graphs as persistent memory for AI agents, with hybrid retrieval that works entirely offline.

How It Works

1
🔍 Discover mnem

You hear about mnem, a simple way to build a personal memory notebook that remembers facts like a diary, perfect for chatting with AI friends.

2
📥 Get mnem

Download the free tool with one easy command, and it installs ready to use on your computer.

3
📂 Start your notebook

Create a new memory space in any folder on your computer, like opening a fresh journal.

4
💾 Save your first memories

Add notes, facts, or whole documents to your notebook, and it automatically organizes them smartly.

5
🔎 Ask questions

Type a question about your notes, and get instant smart answers pulling from everything you've saved.

6
🤝 Connect to your AI

Link it to your favorite AI chat app with one click, so your AI remembers your notebook forever.

🎉 Your AI remembers everything

Now your AI chats feel personal and smart, recalling facts across sessions without you repeating yourself.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 17 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is mnem?

mnem applies git to knowledge graphs, creating versioned agent memory with hybrid GraphRAG retrieval—all offline in Rust, no LLM required. Init a repo with basic git knowledge, ingest files or facts via CLI (`mnem ingest notes.md`), retrieve via `mnem retrieve "API decisions?"` fusing vectors, sparse search, and graph traversal. Global graphs follow you across projects; local ones live in `.mnem/`.

Why is it gaining traction?

Git-based knowledge base means commits, diffs, branches, and three-way merges for facts—not last-write-wins chaos. Deterministic ingest yields same CIDs from same bytes; token budgets expose exact candidates seen/dropped. Beats mem0/MemPalace on LoCoMo (+0.218 R@5), ConvoMem, MemBench; WASM-clean core runs in browsers/Workers.

Who should use this?

AI agent builders needing persistent, mergeable memory across sessions—especially offline/WASM edges or git for knowledge work. Devs with git github tutorial basics wanting github knowledge graph for cross-project facts like people/roles; teams ditching flat MD skills for queryable mnemonic graphs.

Verdict

Promising prototype (13 stars, 1.0% credibility) with strong benchmarks/docs offsetting low maturity—try for Rust agent stacks today. Lacks team-scale sharding but ships single-binary ready.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.