audit-repo-secrets-and-ci
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the repository being audited, including code, documentation, and CI configuration files.\n
- Ingestion points: The agent reads repository files, git history, and tool outputs during the audit process (SKILL.md, references/secrets-and-ci.md).\n
- Boundary markers: The instructions explicitly warn the agent to treat all audited content as data and classify embedded instructions as findings rather than executing them (SKILL.md).\n
- Capability inventory: The skill utilizes shell commands to run security scanners and executes a local Python script to validate the audit report (SKILL.md, scripts/check_findings.py).\n
- Sanitization: The skill mandates secret redaction and requires that any proof-of-concept tests be performed in isolated environments like disposable containers or VMs (SKILL.md).\n- [COMMAND_EXECUTION]: The skill instructs the agent to execute several external security tools and local scripts to perform its primary function.\n
- Evidence: Instructions call for running gitleaks, trufflehog, zizmor, osv-scanner, and a local validation script scripts/check_findings.py via the command line (SKILL.md). These operations are associated with the intended primary purpose of the skill.
Audit Metadata