ci-cd-containerization-advisor
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted configuration data (e.g., Dockerfiles, CI YAML) but acts only as an advisor. Evidence Chain: 1. Ingestion points: Build files, CI workflows, Dockerfiles, and deployment manifests identified in SKILL.md. 2. Boundary markers: Not specified. 3. Capability inventory: The skill is restricted to generating text-based advice and does not possess file-writing or network-execution tools. 4. Sanitization: Not specified. Due to the advisory nature, the risk is negligible.
- [PRIVILEGE_ESCALATION]: The skill includes defensive instructions that recommend running containers as non-root and avoiding privileged access unless strictly necessary.
- [CREDENTIALS_UNSAFE]: The skill promotes secure secret handling by advising on the separation of build-time and runtime secrets and recommending auditable configuration injection over hardcoded values.
Audit Metadata