codex-review

Triggered when user input contains: Running codex review --uncommitted alone only shows AI "what was done (Implementation)". Recording intention first tells AI "what you wanted to do (Intention)". "Code changes + intention description" as combined input is the most effective way to improve AI code review quality. This skill operates in two phases: Decide review mode based on output: Before any review, must check if CHANGELOG.md contains description of current changes. If CHANGELOG is not updated, you must automatically perform the following (don't ask user to do it manually): Auto-generated CHANGELOG entry format: Example - Auto-generation Flow: Before invoking codex review, must add all new files (untracked files) to git staging area, otherwise codex will report P1 error. If there are new files, automatically execute: Explanation: Count change scale: Why use git diff --stat HEAD : Difficulty Assessment Criteria: Model + Reasoning Effort Combinations: Critical Tasks (meets any condition, use best quality model): Difficult Tasks (meets any condition): Normal Tasks (other cases): Evaluation Method: You MUST parse the git diff --stat HEAD output correctly to determine difficulty: Critical: Why the summary line matters: Parsing Rules: Important Edge Cases: Decision Logic (check in order, first match wins): Example Cases: Invoke codex-runner Subtask: Use Task tool to invoke...