gemini-nano-banana-2
Pass
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted user prompts and external image files which constitute an indirect prompt injection attack surface.\n
- Ingestion points: The
--promptand--input-imagearguments inscripts/generate_image.pyingest external content from the user.\n - Boundary markers: No specific delimiters or instructions are used to separate user input from the overall model context.\n
- Capability inventory: The skill can write files to the local system using the
PIL.Image.savemethod inscripts/generate_image.py.\n - Sanitization: No sanitization or filtering is performed on the prompt text or image content before it is processed by the API.\n- [EXTERNAL_DOWNLOADS]: The script declares dependencies on legitimate, well-known Python libraries.\n
- Evidence: The
scripts/generate_image.pyfile specifiesgoogle-genai>=1.0.0andpillow>=10.0.0as requirements, which are fetched during execution.\n- [COMMAND_EXECUTION]: The skill utilizes command-line execution to run its internal processing script.\n - Evidence:
SKILL.mdprovidesuv runcommands to execute thescripts/generate_image.pyscript for its core image generation and editing tasks.
Audit Metadata