offensive-cicd-pipeline
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze external data, such as CI/CD configuration files and source code. This creates an attack surface where instructions hidden within the analyzed data could attempt to mislead the agent.
- Ingestion points: External source code, binary bundles, and CI/CD configuration files (GitHub Actions, GitLab CI, Jenkins) accessed during inventory and audit phases.
- Boundary markers: The instructions do not define explicit delimiters or 'ignore' directives to separate the data being analyzed from the agent's operational instructions.
- Capability inventory: The agent is instructed to perform static analysis, binary disassembly, and vulnerability verification using various search and analysis tools.
- Sanitization: There are no explicit instructions for the agent to sanitize or ignore embedded natural language commands found within the target files.
- [NO_CODE]: The provided skill files consist strictly of Markdown documentation and YAML configuration. No executable scripts (such as Python, JavaScript, or Shell scripts) are included in the package.
Audit Metadata