review
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, such as source code, diffs, and pull requests. This creates a potential surface for indirect prompt injection where instructions embedded in the reviewed code could attempt to influence the agent's output. However, the skill is explicitly instructed to remain read-only and not perform file modifications, which significantly limits the potential impact.
- Ingestion points: The skill accepts diffs, branches, commits, PRs, or file paths as input as defined in the frontmatter of SKILL.md.
- Boundary markers: The instructions direct the agent to "Set the frame" and "review directly," though specific data delimiters for the untrusted content are not explicitly defined in the markdown instructions.
- Capability inventory: The skill's operations are restricted to reading code and generating a textual report. It explicitly prohibits editing files or posting comments, as noted in the "Review" section.
- Sanitization: No explicit sanitization or filtering of the input code is specified.
Audit Metadata