item-selling-point
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/lib/providers.mjsuseschild_process.spawnto execute thedlazycommand-line utility. This is used to interface with the vendor's image generation service and is a standard part of the skill's workflow.\n- [EXTERNAL_DOWNLOADS]: The skill performs legitimate network operations to fetch images from user-provided URLs and download results from reputable AI service providers (OpenAI, Google Gemini, Fal.ai, etc.).\n- [DATA_EXFILTRATION]: The tool is designed to read local image files and transmit them to external AI APIs for processing. This data flow is transparent and inherent to the product's functionality as an AI-driven image generation assistant.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied marketing text and product features to generate image overlays.\n - Ingestion points: User-provided functional points and marketing benefits are passed as parameters to
gen.mjsandbrand.mjs(SKILL.md).\n - Boundary markers: The skill utilizes structured prompt templates and provides explicit negative constraints and safety instructions to the agent to maintain output quality and compliance.\n
- Capability inventory: The skill has the capability to write files to the local system (
scripts/gen.mjs) and make network requests/execute local commands (scripts/lib/providers.mjs).\n - Sanitization: Input is constrained by character limits and specific prompt formatting instructions defined in the documentation.
Audit Metadata