Show HN: Mnemolith – AI coding memory that goes stale when your code changes

Category: ai-ml

Tags: ai-memory, mcp, redis-compatible, vector-search, coding-assistant

Score: 6.5/10 (Innovation: 6, Technical: 7, Documentation: 8, Utility: 5)

Mnemolith is a local, shared memory system for AI coding tools that automatically invalidates memories when linked code files change, using git hashes. It combines a custom Redis-compatible engine, HNSW vector search, and an MCP server to provide provenance, conflict resolution, and sync into AGENTS.md. The idea of tying AI memory to code changes is a novel and practical approach to keeping AI context accurate.

Target audience: AI-assisted developers, backend engineers using Claude Code/Cursor

Repository: https://github.com/anishchugh05/mnemolith · Python · MIT · 1 stars

View on Hacker News