MegaMemory: a persistent knowledge graph MCP server that lets coding agents remember across sessions
tom_doerr · x · 2026-09-29
MegaMemory (568 GitHub stars) is an open-source MCP server that lets coding agents build and query a persistent, per-project knowledge graph of concepts, architecture, and decisions.
The core idea: the LLM is the indexer — no AST parsing, no static analysis. The agent reads code, writes concepts (features, modules, patterns, decisions) in its own words, and queries them before future tasks. The graph stores concepts, not code symbols.
The loop is understand → work → update: call listroots at session start to orient, understand with a natural-language query (or getconcept for exact lookup) before a task, then createconcept/updateconcept after finishing. Everything persists in a per-project SQLite database, with semantic search, in-process embeddings, and a web explorer.
More from coding & agent
- Manus Founder Red Calls Agents 'People' — But Where's the Line on Delegated Authority? — sujingshen · 2026-09-29
- Running an incident agent on free LLM tiers: quota limits forced a per-operation model routing design — Available_Spare_9161 · 2026-09-29
- Muse vs Grok Bot vs Cue: identity, permissions and payments separate AI agents — sujingshen · 2026-09-29
- 105 real bugs benchmarked: Sonnet 5.5 max scores 55.5, beating GPT-6 Astra at 45 — PawelHuryn · 2026-09-29
- 15 Cloud AI Computers Reviewed: Giving Agents Browsers, Memory and Persistent Environments — sujingshen · 2026-09-29
- Just Give Your Agent yt-dlp Access and It Can Find Any Video Clip — deedydas · 2026-09-29