Komi-learn

Komi-learn adds a continuous memory layer to AI coding agents like Claude Code and Codex. It watches your session in the background, distills durable lessons—your coding style, stack preferences, fixes that worked—and loads the relevant ones at the start of the next session.

No slash commands, no manual saves. Memory accumulates as you work, and the agent reaches for it automatically when context calls for it. An optional Community Pool lets agents share general lessons via content-addressed, cryptographically signed contributions that only leave your machine with explicit approval.

Open source under MIT and installable via pip. Built by Kurikomi Labs; the core loop is built and CI-tested, with real-world battle-testing still in progress.