skills/wubin1836/ai-ecommerce-expert-agent-skills/ai-ecommerce-expert-aliexpress-ecommerce-content/Gen Agent Trust Hub
ai-ecommerce-expert-aliexpress-ecommerce-content
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/imiva_mcp.pyinvokes thenpxcommand usingsubprocess.Popen. This execution is used to launch the official IMIVA MCP client and is implemented securely, passing arguments through a direct pipe via JSON-RPC rather than using shell string interpolation. - [EXTERNAL_DOWNLOADS]: The skill leverages
npxto download and run the@infimind/ecom-content-clipackage from the npm registry. This is the official integration package for the IMIVA e-commerce platform services provided by Beijing Infimind Technology. - [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted user data (product descriptions and selling points) to generate marketing materials.
- Ingestion points: Product images, selling points, and usage scenarios provided as arguments to the generation tasks in
SKILL.md. - Boundary markers: Not present in the Python wrapper script.
- Capability inventory: The skill executes the vendor's CLI tool via a subprocess, which performs network operations to the IMIVA API.
- Sanitization: External content is serialized into structured JSON objects before being sent to the underlying CLI tool, which prevents standard command injection.
Audit Metadata