independent-reviewlisted

# Independent review A worker's report is a claim from the worker that wrote the code. When a result matters, someone who did not write it should judge it. Once `review-work`'s gate decides a separate reviewer worker is warranted, this skill creates and briefs it. The reviewer applies `review-task`, the worker wrapper over the shared `review-change` method, and returns findings; `review-work` keeps the verdict and the loop. ## When to use `review-work` owns the gate: it decides whether to review and whether a separate reviewer worker is warranted. This skill applies once that gate picks one. - The change is important or risky — authentication, money, user data, migrations, public interfaces, concurrency, or parallel writers on shared files. - A criterion is hard for the implementer to check honestly — its own assumptions, performance, security, or a subtle edge case. - The developer asks for a second opinion, or the implementer disputes a finding. ## When not to use - The change is trivial and low-risk: review it yourself with `review-change`. Independence is worth a second agent where a wrong result is expensive, not for a one-line config change. - The work is already independently reviewed and only rework remains: see `run-rework`. - Nothing has returned yet: wait with `monitor-agents` first. ## Inputs - The persisted task (`python3 scripts/tm.py task show <id>`) for the goal, every acceptance criterion, and the current iteration. - Fresh evidence —