Mnemo is an open-source memory system for AI coding agents. It captures every session, builds a living archive of knowledge, and lets your agent recall context instantly — across projects, across time. Four tiers. Local-first. Private.

How does the four-tier architecture work?

Tier 0 (Engram) provides instant pattern matching. Tier 1 (Memory DB) stores vector embeddings for semantic search. Tier 2 (Curated Memory) holds extracted knowledge. Tier 3 (Knowledge Graph) maps relationships between entities.

Which AI tools does mnemo support?

Mnemo is a CLI-first tool that works with every AI coding agent. It provides comprehensive hooks for Claude Code, plus MCP server integration with 17 tools.

Yes. Everything stays on your machine. SQLite, Ollama, optional cloud LLM. No accounts. No cloud sync.

Is mnemo free and open source?

Yes. Apache 2.0 licensed. No usage limits, no cloud service.

How is mnemo different from a vector database?

Mnemo adds n-gram pre-filtering, curated knowledge extraction, and a knowledge graph on top of vector search — plus adaptive ranking that learns from usage.

Do I need to change how my agent works?

No. After init, hooks deploy automatically. Your agent works normally. If .mnemo/ doesn't exist, hooks exit silently.

Documentation

GitHub README

Quick start, architecture overview, and CLI commands.

Releases

Version history, changelogs, and download links.

Changelog

Latest releases and version updates.

FAQ

Common questions about Mnemo and how it works.

Docs