code-review
Pass
Audited by Gen Agent Trust Hub on Jul 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructions create a vulnerability surface for indirect prompt injection by telling the agent to prioritize instructions found in the analyzed data over its own primary guidelines.\n
- Ingestion points: The skill processes code modifications, diffs, PRs, and developer messages (
SKILL.md).\n - Boundary markers: Absent. The skill lacks instructions to isolate or ignore malicious instructions embedded in the code being reviewed.\n
- Capability inventory: The skill generates structured JSON suggestions but does not have file system or network access.\n
- Sanitization: Absent. There is no evidence of filtering or validation of the input code changes.\n
- Evidence: In
SKILL.md: "Those guidelines should be considered to override these general instructions" and "defer to any subsequent guidelines that you encounter."
Audit Metadata