The Agent Skills Directory

[SAFE]: The skill utilizes the official google-genai Python library, which is a verified package from a trusted organization (Google).
[SAFE]: Sensitive credentials, such as the GEMINI_API_KEY, are correctly handled through environment variables, avoiding the risk of hardcoded secrets.
[SAFE]: File system operations are localized and limited to reading/writing image assets (e.g., input.jpg, output.png) necessary for the skill's primary function of image generation and editing.
[PROMPT_INJECTION]: The skill has an inherent surface for indirect prompt injection as it processes user-provided text and image data. However, the risk is mitigated as the input is passed to the Gemini API which employs internal safety filters, and the skill does not perform dangerous operations based on this input. 1. Ingestion points: user-provided prompt and input.jpg in SKILL.md. 2. Boundary markers: absent. 3. Capability inventory: Google API network access, local file reading and writing. 4. Sanitization: absent.

gemini-imagegen