dockerize-project

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (LOW): The skill presents an indirect prompt injection surface by reading local files to generate configurations.\n
  • Ingestion points: Files such as package.json, pyproject.toml, and manage.py are analyzed to infer tech stacks.\n
  • Boundary markers: Absent; the skill does not specify delimiters to separate project file content from instructions.\n
  • Capability inventory: The skill generates text code blocks for Docker files but is explicitly forbidden from executing them.\n
  • Sanitization: Absent; the agent infers values directly from file metadata without filtering.\n- [COMMAND_EXECUTION] (SAFE): The skill explicitly forbids the use of sudo and does not perform any automated command execution.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 12:27 PM