Continual Learning
Keep AGENTS.md current by delegating the memory update flow to one subagent.
Trigger
Use when the user asks to mine prior chats, maintain AGENTS.md, or run the continual-learning loop.
Workflow
- Call
agents-memory-updater. - Return the updater result.
Guardrails
- Keep the parent skill orchestration-only.
- Do not mine transcripts or edit files in the parent flow.
- Do not bypass the subagent.