ai-ecommerce-expert-campaign-kv-style-migration

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

• Third-party content exposure detected (medium risk: 0.30). IMIVA 电商专家所需运行时通过 agents/openai.yaml 的 default_prompt 将用户输入（商品素材/卖点/规格/描述等）组织为 MCP 工具调用参数，并在 scripts/imiva_mcp.py 的 invoke() 里经 tools/call 把该自由文本参数直接送入 create_visual_migration_task，因此存在被外部用户自由文本注入的间接提示注入风险。

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

• Potentially malicious external URL detected (high risk: 0.90). scripts/imiva_mcp.py invokes npx to run the remote npm package "@infimind/ecom-content-cli@latest" at runtime (also referenced in SKILL.md and references/mcp-config.example.json), which will fetch and execute remote code that the skill depends on.