STAGE 1RECALL
Shared team memory for every repo.
An MCP server that turns every AI coding session into team knowledge. It logs every decision and why, flags mistakes so nobody repeats them, and briefs the next AI session before the first line of code. Works with Claude Code, Cursor, Codex, Gemini CLI and OpenClaw.
- It listens Every AI session in the repo, from Claude Code, Cursor, Codex or anything that speaks MCP, gets saved when it ends.
- It keeps what matters Decisions and the reason behind them. Mistakes and how to avoid them. What actually shipped. The raw chat gets thrown away.
- It briefs the next one The next session, from anyone on the team, loads that memory first. It knows the rules before it writes a line, so it doesn't drift and doesn't repeat old mistakes.












