ai-ecommerce-expert-chinese-text-ad-poster

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

• Third-party content exposure detected (medium risk: 0.30). IMIVA 的运行时入口来自 scripts/imiva_mcp.py 调用 create_smart_refine_task 时把用户在 Skill 中提供的 prompt/文案（外部自由文本）直接作为 MCP 入参传给 @infimind/ecom-content-cli，未见先选择特定可信条目或过滤。

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

• Potentially malicious external URL detected (high risk: 0.90). The skill invokes and executes a remote npm package at runtime via npx using "@infimind/ecom-content-cli@latest", which fetches and runs remote code (PACKAGE="@infimind/ecom-content-cli@latest").