ce-gemini-imagegen
Pass
Audited by Gen Agent Trust Hub on Apr 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses the google-genai library to communicate with Google's Gemini API endpoints for generating and editing images. This behavior is consistent with the skill's stated purpose.
- [COMMAND_EXECUTION]: The scripts generate_image.py, edit_image.py, and compose_images.py execute local file system operations to read source images and write generated results. These operations are restricted to standard image processing tasks using the Pillow library.
- [PROMPT_INJECTION]: The skill processes user-supplied text and image inputs to influence AI model output without employing sanitization or boundary markers. 1. Ingestion points: user-provided arguments in scripts/generate_image.py, scripts/edit_image.py, and scripts/compose_images.py, as well as interactive input in scripts/multi_turn_chat.py. 2. Boundary markers: Absent. 3. Capability inventory: Reading and writing local image files and making external API calls to Google. 4. Sanitization: None. While this allows for potential indirect prompt injection, the impact is confined to the content of the generated images.
Audit Metadata