ai-ecommerce-expert-multi-platform-content-adaptation

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). 运行时通过脚本 scripts/imiva_mcp.py 调用 IMIVA MCP 工具（如 create_image_expand_task/get_user_tasks）时，会直接把用户在 --args 中提交的自由文本字段（如 sellingPoints、targetAudience、usageScenario 等）作为 JSON 参数传给外部 MCP 并被读取处理。

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill invokes npx -y @infimind/ecom-content-cli@latest at runtime (see scripts/imiva_mcp.py), which fetches and executes a remote npm package that the skill requires, so remote code is executed during runtime: @infimind/ecom-content-cli@latest