The Agent Skills Directory

[SAFE]: The skill contains no malicious instructions or executable code snippets that pose a risk to the host system or data.
[INDIRECT_PROMPT_INJECTION]: The skill handles untrusted data in the form of external code review feedback. It provides strategies to mitigate risks associated with this surface.
Ingestion points: Feedback received from external reviewers or human partners described in SKILL.md.
Boundary markers: The skill instructs the agent to restate requirements in its own words and verify suggestions against the actual codebase before implementation, acting as a logical boundary.
Capability inventory: References the use of grep for codebase verification and gh api for interacting with GitHub pull request comments.
Sanitization: No automated sanitization is described; the skill relies on the agent's technical evaluation and verification steps to filter incorrect or malicious suggestions.
[COMMAND_EXECUTION]: The skill mentions standard development tools like grep and gh api for the purpose of verifying code and responding to review threads, which is consistent with its stated purpose.

receiving-code-review