reviewlisted

# code:review > A PR is open, or a branch has been sitting a while, or you just want to know what's > wrong with the repo. This skill is the one review surface for all three: it recaps a > session and lands every PR it opened (default), it spawns a sub-agent to review one or > more PRs cold with file:line grounding (given PR numbers), or it runs a read-only > architecture-and-quality diagnostic over a whole repo or path (given `repo` / a path / > `--since`). The author never grades their own paper — every PR mode spawns someone who > did not write the code. **When this skill is the default path:** right after `gh pr create`, check whether the repo's automated code reviewer is configured and posting on this PR (see `gh-merge-guard` / `truly-agentic-git-workflow`). If the bot is working, wait for it — do not stack this skill on top. If the bot is missing, silent, down, or the repo has none, run this skill **immediately** so merge-on-green is not blocked. Never hand the merge to the user because the automated reviewer is down. ## Anti-bias rule If you (the orchestrator) opened the PR(s) in the current session — and most of the time you did — you **must not** be the reviewer. Spawn a sub-agent. The whole point is fresh eyes. This applies to every PR-touching mode below; the whole-repo mode has no PR to be biased about. ## Scope resolution — pick the mode `$ARGUMENTS` parses into one of three modes: | Pattern | Mode | What happens | |---|---|---| | empty, or `dry-run` / `no