structural-review
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process untrusted data (PR diffs), which creates a vulnerability surface for indirect prompt injection where malicious instructions inside the code diff could attempt to manipulate the review output.
- Ingestion points: The skill ingests PR diffs as the primary input for its structural review logic (SKILL.md).
- Boundary markers: There are no explicit instructions for the agent to use delimiters or ignore embedded instructions when parsing the external code content.
- Capability inventory: The skill is explicitly restricted to a 'Report-only' role, stating it does not mutate files or perform network operations, which mitigates the risk of the agent performing harmful actions if an injection occurs.
- Sanitization: The rubric does not include steps for sanitizing or validating the code input to prevent the interpretation of embedded natural language instructions.
Audit Metadata