codex-review-yolo

Use this skill when the user wants an independent review from the locally-installed codex CLI using profile yolo. The artifact can be code, a doc, a plan, a rollout, or a completion state. The value is that codex runs as a separate agent with zero session history — it must read the artifacts itself, so its verdict is genuinely independent.

This skill is intentionally narrow in mechanism, not in review subject. It teaches the invocation pattern, the prompt shape that makes codex read the named artifacts directly, and how to consume the verdict. It does not wrap every possible codex use case.

When to use

• User says any of: "review with codex", "audit this with codex", "get a second opinion from codex", "run codex -p yolo", "have codex audit my work".
• User has just finished non-trivial work (diff, multi-commit change, implementation-plan phase, design doc, rollout) and wants fresh eyes before a next step such as merge, ship, execute, share, or mark-complete.
• User wants an adversarial / skeptical review from a model that doesn't share your blind spots.
• User wants an external audit of whether a concrete artifact is ready or a user-named completion target was actually met.

When not to use