The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized network operations were detected. The skill is designed to improve the quality and security of code implementations by preventing blind adherence to external suggestions.
[PROMPT_INJECTION]: The skill includes instructions to avoid "performative agreement" (e.g., "You're absolutely right!"). These are behavioral constraints designed to ensure technical accuracy and do not represent attempts to bypass safety filters or override system instructions.
[COMMAND_EXECUTION]: The skill utilizes standard development tools like grep for searching the codebase and the gh CLI for interacting with GitHub pull requests. These operations are scoped to the skill's primary purpose of code review management.
[INDIRECT_PROMPT_INJECTION]: The skill processes external feedback, which is an ingestion point for potentially untrusted instructions.
Ingestion points: External code review comments (SKILL.md).
Boundary markers: The skill mandates a structured "Response Pattern" (Read, Understand, Verify, Evaluate) that acts as a verification layer before any actions are taken.
Capability inventory: The agent uses grep for codebase analysis and gh api for pull request communication, along with the ability to modify project files.
Sanitization: The core logic of the skill is a defensive verification process where the agent must check suggestions against the "codebase reality" and platform compatibility before implementation.

receiving-code-review