The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard development CLI tools to verify work completion and technical correctness. Verification gates include commands such as bun test, npm test, bun run build, and bunx tsc --noEmit. These are intended to provide evidence of correctness before the agent claims a task is complete.
[COMMAND_EXECUTION]: Version control operations are used to manage the code review process, specifically calling git rev-parse HEAD and git log to identify commit SHAs for review by a subagent.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external code review feedback, which represents a surface for indirect prompt injection where an attacker might embed malicious instructions in a pull request or code comment.
Ingestion points: External feedback is received from third-party reviewers as described in SKILL.md and references/code-review-reception.md.
Boundary markers: The instructions implement a mandatory response pattern (READ → UNDERSTAND → VERIFY → EVALUATE → RESPOND → IMPLEMENT) and explicitly warn the agent that external feedback should be treated as suggestions to evaluate, not orders to follow.
Capability inventory: The skill uses tools for git operations, testing, linting, and building (git, bun, npm, tsc).
Sanitization: The protocol requires the agent to verify all suggestions against the current codebase and provides instructions for pushing back against technically incorrect or contextually inappropriate feedback, acting as a logical filter for external instructions.

code-review