dockerfile-best-practices
Pass
Audited by Gen Agent Trust Hub on May 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a comprehensive security-focused documentation and template repository for Docker development. Key findings include:
- [PROMPT_INJECTION]: No attempts to override agent behavior or bypass safety filters were found. The instructions are purely task-oriented.
- [DATA_EXFILTRATION]: No exfiltration patterns detected. The skill specifically includes guidance and '.dockerignore' templates to prevent accidental credential exposure and data leakage.
- [OBFUSCATION]: All content is in plain-text markdown. No Base64, zero-width characters, or hidden payloads were identified.
- [REMOTE_CODE_EXECUTION]: Dockerfile examples use standard package managers (apt, npm, pip) and official base images from well-known registries (gcr.io, Docker Hub). No untrusted remote script execution patterns were found.
- [PRIVILEGE_ESCALATION]: The skill explicitly teaches against privilege escalation by mandating the use of non-root users in the runtime stage and mentioning 'no-new-privileges' security options.
- [DYNAMIC_CONTEXT_INJECTION]: No use of shell-execution placeholders or '!' command syntax was found in the SKILL.md or reference files.
Audit Metadata