ai-image-generator
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses
curlto fetch model information from Google's official Gemini API endpoint (generativelanguage.googleapis.com). This is a well-known service used here for valid model discovery. - [COMMAND_EXECUTION]: Provides shell commands for environment variable configuration and image format conversion (WebP, cropping). These are standard development tasks.
- [DYNAMIC_EXECUTION]: Employs Python heredocs to manage API requests and process binary image data. The scripts are static templates included in the skill and do not execute untrusted remote code; the piping of API output to Python is used solely for JSON parsing of model names.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text to form API payloads. It mitigates injection risks by using
json.dumpsfor proper serialization and provides clear boundary guidance for the AI through a structured 5-part prompting framework. - Ingestion points: User prompts interpolated into Python scripts in
SKILL.md. - Boundary markers: The skill uses a structured framework to clearly separate image components and constraints.
- Capability inventory: Network requests (urllib), file writing (open/wb), and bash execution.
- Sanitization: Employs standard Python JSON libraries to escape user input before transmission.
Audit Metadata