qol-code-reviewlisted

# Code Review ## Overview Run independent reviewer agents over the same change set, keep findings separated by domain, then synthesize a severity-ordered go/no-go. The skill is for review orchestration, not implementation, unless the user explicitly asks to apply fixes after the review. Default safety posture: read-only. Do not edit files or write comments unless the user explicitly requests edits. Always persist the finished review to a temporary markdown file via the bundled writer (see "Persisting the review"). This applies to every run, board or solo, pass or block. Writing that one temp artifact is the sole exception to the read-only posture; it never touches the reviewed codebase. ## Board Mode Contract - Treat an explicit `$qol-code-review`, `$qol-code-review:code-review`, or "use the qol-code-review skill" invocation as review-board mode unless the user explicitly asks for a solo/no-agent review. - Spawn agents for review-board mode. Do not perform a parent-thread solo review and present it as a qol-code-review skill result. - In Codex, explicit asks include phrases like "spawn agents", "review board", "delegate reviewers", "parallel reviewers", "one agent per point", naming multiple reviewer agents to run, or invoking the qol-code-review skill by name. - In Codex review-board mode, read `references/codex-adapter.md` before reviewer selection and obey it as mandatory execution mechanics. - If Codex multi-agent tools are not visible, use `tool_search` with `mult