skills/wubin1836/ai-ecommerce-expert-agent-skills/ai-ecommerce-expert-xiaohongshu-note-package/Gen Agent Trust Hub
ai-ecommerce-expert-xiaohongshu-note-package
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions and the helper script
scripts/imiva_mcp.pyutilizenpxto download and run the@infimind/ecom-content-cli@latestpackage. This fetches the most recent version of the code from the public NPM registry at runtime. - [COMMAND_EXECUTION]: The script
scripts/imiva_mcp.pyuses thesubprocess.Popenfunction to launch thenpxcommand. This facilitates the execution of the external MCP tool within the local environment. - [REMOTE_CODE_EXECUTION]: By invoking
npx -y @infimind/ecom-content-cli@latest, the skill performs remote code execution by fetching and executing a third-party binary package directly from a remote registry. - [PROMPT_INJECTION]: The skill acts as an interface for generating social media notes by ingesting user-provided data including image paths, product names, and marketing copy, creating a surface for indirect prompt injection.
- Ingestion points: Data enters the system through the
argsparameter in thecallcommand, including fields likeimages,productName, andsellingPointsinscripts/imiva_mcp.py. - Boundary markers: The skill does not implement specific delimiters or explicit instructions to the model to ignore potential commands embedded within the product data.
- Capability inventory: The skill possesses the capability to execute shell commands via
subprocess.Popeninscripts/imiva_mcp.py. - Sanitization: Input data is converted to JSON and passed to the external tool without visible sanitization or validation of the content.
Audit Metadata