What problem does it solve? Multi-stage AI agent workflows make many decisions (tier selection, skill skips, critic verdicts, scope changes) that are lost when the conversation ends. This Skill maintains a persistent audit trail so you can answer "why did the orchestrator make this choice?" without the original transcript. ## Core Features & Use Cases - Decision Logging: Records Control Tower decisions (tier, skips, topology, critic verdict, outcome) to memory_bank/orchestrator-log.md after preflight, critic review, scope changes, Hard Stops, and closeout. - Subagent Result Tracking: Logs each subagent's verdict, key findings, and evidence summary to memory_bank/review-log.md after every return. - External Team Consolidation: Reads external-team-log.md from delegated Claude Code handoff sessions and consolidates verdicts into the review log during closeout. - Use Case: After a Work Block completes, an Owner reviews orchestrator-log.md to see why Standard tier was chosen, which skills were skipped, and what the critic verdict was — all in one table row. ## Quick Start Ask the agent to log the current orchestrator decision and subagent results to the session audit trail.
