The Agent Skills Directory

[SAFE]: The skill provides instructional material and code samples for integrating with Google Gemini's image generation capabilities using the official google-genai SDK from a well-known service.
[SAFE]: Recommended security practices are followed by advising users to store API keys in environment variables rather than hardcoding them in scripts.
[SAFE]: All library dependencies referenced, such as google-genai and pillow, are standard, reputable libraries for AI interaction and image processing.
[PROMPT_INJECTION]: The provided code patterns demonstrate how to process untrusted user data (text prompts and reference images), which represents a surface for indirect prompt injection.
Ingestion points: Python functions in references/python.md accept prompt and reference_image_path parameters.
Boundary markers: Boundary markers or explicit 'ignore embedded instructions' warnings are absent from the example code snippets.
Capability inventory: The documented patterns include network access via the SDK and file system write access for saving generated results.
Sanitization: Input validation and content escaping are not demonstrated in the reference implementation.

gemini-image