article-batch-illustration
Pass
Audited by Gen Agent Trust Hub on Apr 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script (
scripts/generate_image.py) to manage API requests and save images. This is a standard and safe implementation for agent tasks that requires complex logic or specific API handling. - [EXTERNAL_DOWNLOADS]: The skill communicates with Google's Gemini API at
https://generativelanguage.googleapis.com. This is a well-known service and the interaction is necessary for the skill's primary function of image generation. - [PROMPT_INJECTION]: The skill processes untrusted article content to generate image prompts, creating a surface for indirect prompt injection. (1) Ingestion points: Article content is read from the user's environment in
SKILL.md. (2) Boundary markers: No specific delimiters or instructions are used to isolate the untrusted article text during prompt construction. (3) Capability inventory: Thegenerate_image.pyscript has permissions to write files to the local file system and initiate network requests to the Gemini API. (4) Sanitization: No input validation or sanitization of the article content is performed before it is incorporated into prompts.
Audit Metadata