threejs-image-generator
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill follows secure practices by instructing users not to hardcode API keys and providing scripts that read from environment variables or command-line arguments. All external dependencies (google-genai, pillow) are well-known libraries from trusted organizations.
- [COMMAND_EXECUTION]: The documentation references a credential probing script (probe_asset_credentials.sh) located in a related skill directory. This is used for local environment configuration and does not indicate malicious activity.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text and images as inputs for an AI model, which is the primary function of the tool. While this represents a standard attack surface for indirect prompt injection, it is consistent with the skill's purpose.
- Ingestion points: User-provided strings via --prompt and image files via --input-image in scripts/generate_image.py.
- Boundary markers: The skill does not use specific delimiters or boundary markers when passing the user prompt to the Gemini API.
- Capability inventory: The script has capabilities to read files (PILImage.open), write files to the project directory (image.save), and perform network operations to communicate with Google's GenAI services.
- Sanitization: There is no explicit sanitization or filtering of the prompt text or image content before processing.
Audit Metadata