PROJECTMEM: A Local-First, Event-Sourced Memory and Judgment Layer for AI Coding Agents

In three linesProjectMem is a local-first, open-source memory layer for AI coding agents based on an immutable event log (issues, attempts, fixes, decisions). It reduces token consumption (5,000-20,000 per session) and adds preventive governance: agents are warned before repeating failed fixes or editing fragile files. Runs fully offline via MCP.

Summary generated by Claude — human-verified