check-pr
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from external PR comments and review discussions to identify issues and suggest code modifications.
- Ingestion points: In
SKILL.md, the skill fetches comments and descriptions from GitHub, GitLab, and Perforce usinggh api,glab api, andp4(Step 2 and 5). - Boundary markers: External content is not wrapped in delimiters or preceded by instructions to the agent to ignore embedded commands or instructions.
- Capability inventory: The skill has extensive capabilities including the ability to write to files, commit and push changes to remote repositories, and perform API mutations to resolve review threads (Step 8 and 9).
- Sanitization: No sanitization or validation of the ingested comment content is performed before the agent processes it to determine "Actionable" items.
Audit Metadata