skills/wubin1836/ai-ecommerce-expert-agent-skills/ai-ecommerce-expert-sku-variant-main-image/Gen Agent Trust Hub
ai-ecommerce-expert-sku-variant-main-image
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill uses npx to download and execute the latest version of the @infimind/ecom-content-cli package from the NPM registry. This occurs during skill initialization and tool invocation.
- [COMMAND_EXECUTION]: The helper script scripts/imiva_mcp.py uses the subprocess module to spawn and manage the lifecycle of the MCP server process.
- [EXTERNAL_DOWNLOADS]: The skill relies on fetching an external Node.js package at runtime to provide its core functionality, introducing a dependency on the availability and integrity of the NPM registry and the package maintainer.
- [DATA_EXFILTRATION]: User-provided product information is sent to the external domain imiva.ecpro.com. While this is the intended service endpoint, it represents a data transmission to a non-whitelisted external destination.
- [PROMPT_INJECTION]: The skill processes user-supplied strings for product categories and selling points without explicit sanitization or boundary markers. This creates a surface for indirect prompt injection where malicious input could influence the behavior of the downstream image generation service.
- Ingestion points: Arguments in the create_product_main_image_task tool call found in SKILL.md examples and processed by scripts/imiva_mcp.py.
- Boundary markers: None present in the instructions to isolate user input from system prompts.
- Capability inventory: The skill has command execution and network access capabilities.
- Sanitization: No sanitization is performed on input strings before transmission.
Audit Metadata