The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, specifically pull request descriptions, commit messages, and source code. This creates an attack surface where malicious instructions embedded in the code or comments could attempt to influence the agent's behavior.
Ingestion points: The skill reads pull request descriptions, linked issues, commit history, and complete code diffs as described in the mindset and context sections of SKILL.md.
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to ignore potentially malicious directions embedded within the code or comments it is reviewing.
Capability inventory: The skill uses shell commands to interact with the repository and has broad read access to the local codebase.
Sanitization: No specific sanitization, validation, or filtering logic is defined for the external data before it is processed by the AI model.
[COMMAND_EXECUTION]: The skill provides templates for shell commands using the GitHub CLI (gh) and a suggested extension (gh-pr-review). These commands allow the agent to start reviews, add comments, and submit review statuses.
Evidence: SKILL.md contains examples of gh pr-review review --add-comment and gh pr-review review --submit to facilitate the review process.

critical-code-reviewer