This is a pure orchestration layer that keeps your AGENTS.md file up to date by handing off all the real work to a subagent called agents-memory-updater. You invoke it when you want to mine past conversation transcripts and fold learnings back into your agent memory. The entire parent skill is just a passthrough that calls the updater and returns its result, which is an interesting pattern if you're building multi-agent systems where you want clean separation between orchestration and execution. The guardrails explicitly prevent you from doing any actual file editing or transcript mining at this level, forcing everything through the subagent boundary.

Install to Claude Code

npx -y skills add cursor/plugins --skill continual-learning --agent claude-code