The Agent Skills Directory

[SAFE]: No malicious patterns or behaviors were detected in the prompt instructions or the provided Python script. The skill facilitates image generation and editing through official Google GenAI services.\n- [COMMAND_EXECUTION]: The skill uses uv run to execute a local script for its intended function, which is a standard pattern for AI agent skills and does not involve execution of untrusted commands.\n- [EXTERNAL_DOWNLOADS]: Dependencies (google-genai, pillow) are managed via uv from official registries. These are well-known and trusted libraries for interacting with Google APIs and processing images.\n- [DATA_EXFILTRATION]: Data sent to external endpoints is limited to the user's prompt and image data being sent to Google's official Gemini API, which is the core purpose of the skill.\n- [PROMPT_INJECTION]: While the skill ingests user-provided prompts and images that are then processed by a model, this represents a standard operational surface for multimodal tools. No evidence of malicious exploitation was found. Ingestion points: args.input_image and args.prompt in scripts/generate_image.py. Capabilities: local file read/write and network access to Google APIs. Sanitization: handled by the underlying Gemini API and standard library file handling.

image-nano-banana