mnemon

Memory 604 stars Resource

LLM-supervised persistent memory for AI agents — graph-based recall and knowledge management.

View project Visit website

GitHub stars 604

LicenseApache-2.0

Primary languageGo

Last push2026-09-30T13:22:45Z

Maintainermnemon-dev

Added to this directory2026-08-15

About this project

LLM-supervised persistent memory for AI agents — graph-based recall, cross-session knowledge, single binary. Works with DeepSeek Harness, Claude Code, OpenClaw, and any agent runtime.

More about this project

Zero user-side operation — install once; supported runtimes can use hooks, minimal runtimes can use persistent rules - LLM-supervised — the host LLM decides what to remember, update, and forget; no embedded LLM, no API keys - Multi-framework support — Claude Code, Codex, Cursor, ZCode, TRAE/TRAE Work, Qoder/QoderWork, CodeBuddy, WorkBuddy, Kimi Code, OpenCode, and Hermes Agent (hooks/plugins), OpenClaw (plugins), Pi (extensions), MiniMax Code and Nanobot (skills), DeepSeek Harness (via the dsh-mnemon plugin), and more - Runtime-native integration — runtime-specific SKILL.md, shared guide.md, and supported hooks or extensions - Four-graph architecture — temporal, entity, causal, and semantic edges, not just vector similarity - Intent-native protocol — three primitives (remember, link, recall) map to the LLM's cognitive vocabulary, not database syntax; structured JSON output with signal transparency - Intent-aware recall — graph traversal + optional vector search (RRF fusion), enabled by default for all queries - Built-in deduplication — remember and import skip exact content repeats and preserve distinct facts; similarity suggestions guide review - Retention lifecycle — importance decay, access-count boosting, and garbage collection - Privacy-safe receipts — export hashed operation receipts for memory-boundary audits without raw memory contents or queries - Optional…

Use cases

mnemon is LLM-supervised persistent memory for AI agents — memory managed by a model, not by hand-written rules.

Structured recall. It uses graph-based memory, so recall isn't keyword matching — the agent retrieves related facts the way you'd follow connections in your own notes. This matters for complex projects where context is relational, not flat.

Long-running work. For agents that work on the same codebase for weeks, graph memory accumulates real understanding instead of a growing flat list that's increasingly ignored.

Knowledge management. It doubles as a personal knowledge layer: projects, decisions, rationale — the kind of thing that dies when a teammate leaves or a session ends.

Practical tip: treat it as a system of record, not a cache. Feed it decisions and their reasons, and retrieval quality stays high because the graph keeps connections meaningful. Pair with distill if you want conversation history distilled into the same store.

Project FAQ: mnemon

What is mnemon?

mnemon is a DeepSeek Harness ecosystem resource maintained by mnemon-dev. LLM-supervised persistent memory for AI agents — graph-based recall and knowledge management.

Where can I get mnemon?

Source code and usage instructions are available at https://github.com/mnemon-dev/mnemon. Follow the repository README for the correct setup steps.

Is mnemon free?

mnemon is a community open-source project released under the Apache-2.0 license. Review the repository license and documentation before use.