docker
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [PROMPT_INJECTION] (SAFE): No instructions to bypass safety or override agent behavior were found. The guidelines are strictly technical and educational.
- [CREDENTIALS_UNSAFE] (SAFE): The skill promotes secure secret management and avoids hardcoding credentials, recommending runtime injection instead.
- [COMMAND_EXECUTION] (SAFE): Instructions are limited to standard, non-malicious Dockerfile syntax. No arbitrary shell execution was detected.
- [EXTERNAL_DOWNLOADS] (SAFE): Example Dockerfiles use pinned, official images from trusted repositories (e.g., node:20-alpine).
- [INDIRECT_PROMPT_INJECTION] (LOW): The skill recognizes the risk of injection through build-time variables (ARG/ENV) and includes rules for sanitization. Ingestion point: external build arguments. Boundary markers: absent. Capability: Dockerfile review. Sanitization: recommended in rule text.
- [NO_CODE] (SAFE): The skill is entirely documentation-based with no scripts or binary files.
Audit Metadata