Receiving Code Review gives your agent a six-step response pattern: read the full feedback, restate it, check it against the codebase, evaluate it for this project, respond with a technical acknowledgment or reasoned pushback, and implement one item at a time with a test after each.
What You Get
- A response pattern that runs from reading the feedback to implementing it
- Source-specific handling for your own feedback and for external reviewers
- Guidance on when and how to push back, and how to correct yourself when the pushback was wrong
- An implementation order: blocking issues, simple fixes, then complex ones
Common Use Cases
- Working through a pull request review with many comments
- Evaluating suggestions from an external reviewer or bot
- Handling a request like fix items 1 to 6 when some items are unclear
What Makes It Unique
Most review skills help produce a review. This one governs how an agent receives one. It forbids reflexive agreement, requires verification against the real code before any change, and lists concrete phrases to avoid, so each piece of feedback is evaluated before it is applied.
Under The Hood
The skill is instructions only, with no scripts to run. It shapes how the agent reasons and replies, and it sends the agent to the codebase with search and test commands to verify suggestions. For feedback left on GitHub it tells the agent to reply inside the comment thread through the gh command-line tool, which calls the GitHub API.