aif-dockerize
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: Risk of indirect prompt injection via .ai-factory/skill-context/aif-dockerize/SKILL.md. Evidence: (1) Ingestion point: .ai-factory/skill-context/aif-dockerize/SKILL.md. (2) Boundary markers: Absent. (3) Capability inventory: File writing and Docker command execution. (4) Sanitization: Absent. Malicious context files could manipulate the resulting Dockerfiles and deployment scripts by injecting instructions into the templates.\n- [COMMAND_EXECUTION]: The skill executes git and docker CLI tools for project analysis and container management. It also generates several operational shell scripts (deploy.sh, update.sh, rollback.sh, etc.) and performs chmod +x to enable production workflows.\n- [EXTERNAL_DOWNLOADS]: The skill uses WebSearch and WebFetch to retrieve stable version tags for infrastructure images from official documentation (e.g., angie.software). These are trusted services used to ensure the generation of valid and up-to-date image references.
Audit Metadata