receiving-code-review
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external feedback from human partners and external reviewers, which represents an ingestion surface for untrusted data.
- Ingestion points: Code review feedback provided via GitHub comments or other communication channels (SKILL.md).
- Boundary markers: The skill implements a strict "Response Pattern" (Read, Understand, Verify, Evaluate, Respond, Implement) that requires the agent to verify instructions against the codebase before taking action. It explicitly forbids "performative agreement" (blindly following instructions).
- Capability inventory: Access to terminal execution via a proxy (e.g.,
rtk npm test), filesystem access for documentation (MemoryStore), and GitHub API interactions (gh api). - Sanitization: The instructions include a mandatory "Security & Multi-Agent Hygiene" protocol requiring the redaction of secrets and PII before sharing context with subagents.
- [COMMAND_EXECUTION]: The skill references the use of an
ExecutionProxy Interface(e.g.,rtk npm test) and GitHub API commands (gh api) for replying to pull request comments. These are documented as part of the skill's technical verification and response workflow. - [DATA_EXPOSURE_AND_EXFILTRATION]: The skill framework includes specific protocols for handling sensitive data, emphasizing least privilege and the quarantine of experimental code. It references internal architectural maps (
docs/graph.json) for dependency discovery.
Audit Metadata