review-pr
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code changes from external contributors, creating a surface for indirect prompt injection.
- Ingestion points: Pull request content is fetched using
gh pr diffandgh pr viewin SKILL.md. - Boundary markers: There are no explicit delimiters, XML tags, or specific instructions to the model to treat the retrieved diff content strictly as data rather than instructions.
- Capability inventory: The skill uses the GitHub CLI (
gh) to view diffs and create review comments on the pull request. - Sanitization: The raw diff output is not sanitized, escaped, or checked for embedded adversarial code comments before being processed by the AI.
Audit Metadata