prcheckloop
Pass
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to identify and execute shell commands found within repository workflow files (e.g., .github/workflows/) to reproduce CI failures. This creates a risk where malicious commands embedded in the codebase or workflow configurations could be executed by the agent.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests and acts upon data from external sources like GitHub Action logs.
- Ingestion points: Technical logs retrieved via
gh run view --log-failedand metadata fromgh apicalls (SKILL.md steps 3 and 5). - Boundary markers: The instructions do not define clear boundaries or 'ignore' directives for content parsed from logs.
- Capability inventory: The agent can modify files, execute arbitrary shell commands for reproduction, and perform
git pushto remote repositories. - Sanitization: There is no evidence of sanitization or validation logic to filter malicious instructions that might be embedded in the output of failing tests or build steps.
Audit Metadata