The Agent Skills Directory

[SAFE]: The skill uses official libraries (google-genai, Pillow) to interact with the Gemini API and process images locally. Network operations are restricted to authorized API calls.
[SAFE]: No hardcoded credentials were found. The skill correctly instructs the user to manage secrets like GEMINI_API_KEY via environment variables or CLI flags.
[SAFE]: Python scripts use standard, safe parsing methods. The custom YAML parser in art_spec.py is implemented using regular expressions and string splitting without resorting to dangerous functions like eval() or exec().
[SAFE]: The skill implements automated QA gates (validate_sprite.py, critique_image.py) that run locally to ensure asset quality and consistency, reducing the risk of importing malicious or broken content into the Unity engine.
[SAFE]: Instructions for installing dependencies correctly utilize virtual environments (venv), following modern development standards to avoid system-level tampering.
[SAFE]: Indirect prompt injection risk is mitigated by the use of explicit boundary markers and "style contracts" when building prompts for the image generation model.

unity-image-generator