receiving-code-review

The receiving-code-review skill provides a structured, high-rigor protocol for AI coding agents tasked with handling pull requests or inline feedback. It enforces a strict boundary between performative social pleasantries and actionable technical evaluation, ensuring that code changes are scrutinized against the existing codebase rather than accepted blindly. This skill is essential for maintaining architectural integrity in complex projects where external suggestions or partner feedback must be vetted for technical accuracy, security implications, and potential regressions. By adopting a methodical approach—read, understand, verify, evaluate, respond, and implement—the agent avoids common pitfalls such as partial implementation, batching without testing, and unnecessary feature creep.

• Enforces a rigorous six-step response pattern: Read, Restate, Verify, Evaluate, Respond, and Implement.

• Prohibits performative language like 'You're right' or 'Great point' to ensure professional, technical focus.

• Implements systematic YAGNI (You Aren't Gonna Need It) checks when reviewers suggest unnecessary features.

• Requires verification against the current repository state before acting on any feedback.

• Includes specific conflict resolution paths for human partner versus external reviewer input.

• Standardizes push-back mechanisms using technical reasoning rather than emotional or defensive justifications.

• Facilitates granular, single-item implementation loops with mandatory per-item testing and validation.

• Automatically prompts for clarification when feedback is ambiguous or incomplete, preventing wrong implementations.

• This skill should be active whenever a Pull Request, issue comment, or inline review is received.

• Input expected is any structured review feedback (GitHub/GitLab comments, human partner notes).

• The agent will output either clarifying questions, technical pushback based on codebase analysis, or verified implementation commits.

• Constraints include absolute adherence to technical correctness over comfort, forbidding any expressions of gratitude or apology to maintain pure communication.

• Use this skill to integrate external feedback into existing worktrees without breaking compatibility with established architectural decisions.