pentest-supply-chain
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for ingesting and analyzing untrusted data from repository manifests and CI/CD configuration files.
- Ingestion points: The skill reads external data from package.json, requirements.txt, go.mod, and GitHub Actions workflow files (SKILL.md, references/workflows.md).
- Boundary markers: There are no explicit instructions for the agent to use delimiters or to disregard commands embedded within the data being analyzed.
- Capability inventory: The skill utilizes several diagnostic tools including snyk, npm audit, trivy, and semgrep (references/tools.md).
- Sanitization: The workflow does not describe any sanitization or validation steps to ensure that content from analyzed files does not influence the agent's behavior.
Audit Metadata