The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and implement changes based on external code review feedback and PR notes. This represents a potential surface for indirect prompt injection, as malicious instructions hidden in review comments could attempt to influence the agent's actions. However, this risk is inherent to the intended purpose of the skill. * Ingestion points: Reviewer comments, PR notes, and lists of requested changes provided by the user (SKILL.md). * Boundary markers: None explicitly mentioned for isolating feedback from instructions. * Capability inventory: File system operations via Grep, Glob, Edit, and Write tools, and shell command execution via Ruff. * Sanitization: No explicit validation or filtering of input feedback is described.
[COMMAND_EXECUTION]: The skill utilizes 'uv run ruff check' to validate code changes. This is a standard invocation of a well-known linting tool for code quality and does not represent a security risk in this context.

review-implementing