What problem does it solve? Long-running, autonomous, or multi-phase work leaves no reviewable trail, so a human who steps away cannot verify what was decided, why, or whether the evidence backs it up. ## Core Features & Use Cases - Structured decision log: One TSV row per decision with timestamp, phase, decision, why, evidence, and result columns, kept single-line and spreadsheet-safe. - Helper script: scripts/log.sh appends well-formed rows, stamps ISO8601 timestamps, writes the header on first use, and neutralizes spreadsheet formula injection in cells. - Audit and cross-model review: End-of-run checks reconcile the log against the actual transcript, and a subagent on a different model family flags weak evidence or risky choices. - Use Case: During an unattended multi-hour refactor, each fork, revert, and verification result is logged with a commit SHA or test output, so a reviewer can later read the TSV top to bottom and trust the outcome. ## Quick Start Ask the agent to start a show-me-your-work decision log for this task and record each major decision with its evidence as the run progresses.
