ai-ecommerce-expert-smart-product-retouch

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

• Third-party content exposure detected (medium risk: 0.30). IMIVA Ecommerce MCP is invoked via the user-supplied --args payload (including free-text prompt and images URLs/paths) and sent at runtime to create_smart_refine_task through scripts/imiva_mcp.py, so outsider-authored text can be injected into the LLM/tool workflow without pre-selecting a trusted item.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill invokes npx to fetch and execute the npm package @infimind/ecom-content-cli@latest at runtime (e.g., subprocess [npx, "-y", PACKAGE]), which downloads and runs remote code required by the skill.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). This skill includes explicit MCP calls that query credits and create tasks which deduct platform credits (billing). It documents a get_user_credits call, a dryRun/estimatedCredits flow, and a create_smart_refine_task call that may "扣除积分" and accepts a maxCredits parameter to cap spending. Those are concrete APIs that can initiate or control monetary consumption on the platform (billing/credits), so it grants direct financial execution authority.