What problem does it solve?
Capture end-of-session context, timeline, and lessons so work sessions do not lose momentum or the learnings created during coding, research, or design work. The skill automates assembly of a concise retrospective, an AI diary, and a lesson file while weaving optional pulse context and git metadata into a single narrative for future reference.
Core Features & Use Cases
- Quick Retro: Generate a short, practical retrospective and commit it to the project's memory store.
- Full Template: Produce a detailed session report with timeline, files modified, architecture notes, honest feedback, and metrics for in-depth reflection.
- Timeline Reconstruction: Use session logs and exported .jsonl traces to rebuild session timelines and map activities to commits.
- Deep Analysis: Optionally run a five-agent deep mode to parallelize git analysis, file change summaries, pattern extraction, and Oracle memory searches.
- Persistent Learning: Save lessons to the local Oracle memory and optionally sync insights back to the Oracle using the oracle_learn call.
Quick Start
Run /rrr to create a concise retrospective and save it to psi/memory/retrospectives while also writing a lesson file to psi/memory/learnings and syncing insights to the Oracle.