reflect

What problem does it solve? Valuable lessons from an AI coding session are lost once the conversation ends. This Skill mines the active transcript for durable learnings and turns them into concrete, reviewed edits to existing skills so future agents benefit. ## Core Features & Use Cases - Parallel Multi-Lens Review: Spawns three reviewer subagents (judgment, tooling, divergent) over the session transcript, each with a dedicated prompt template from the references directory. - Structured Synthesis: A synthesizer subagent merges findings into an Accepted / Rejected / Backlog list with explicit routing to skill edits, description tuning, or new skill creation. - Approval-Gated Application: Presents the full synthesis to the user before applying any change, then routes trivial edits directly and substantive edits to the create-skill workflow. - Use Case: After a long debugging session where the agent discovered a CI flag and a missed skill trigger, say "reflect" to extract those learnings and encode them into the relevant skills. ## Quick Start Say "reflect" or "/reflect" at the end of a session to review the transcript and apply approved learnings to your skills.