The Agent Skills Directory

Technical Verification of External Input: The skill establishes a "Response Pattern" that requires the agent to verify suggestions against the codebase reality before implementation. This approach ensures that any external input, such as code review comments, is subjected to technical evaluation, which helps prevent the introduction of errors or suboptimal code.
Mitigation of Indirect Prompt Injection Surface: Code review feedback is a form of external, untrusted data that could potentially contain malicious or misleading instructions. The skill's instructions to "Verify before implementing" and "Push back with technical reasoning" act as functional safeguards. By discouraging "blind implementation" and "performative agreement," the skill reduces the risk of the agent executing instructions embedded in review comments without proper validation.
Standard Tool Usage: The skill references the use of common development tools such as grep for codebase searching and the GitHub CLI (gh api) for managing pull request comments. These tools are used appropriately within the context of a software development workflow.
Behavioral Alignment: The skill includes specific behavioral constraints, such as avoiding expressions of gratitude or blind agreement. These are designed to maintain a professional, technically-focused persona for the agent and do not present security risks.

receiving-code-review