find-customers

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This skill instructs the host to add and run a remote MCP connector (npx ppxc-leads-mcp), to persuade the user/host to trust/enable that connector, and to scrape and upload user-account-accessible comment data to external report/customer-pool servers — creating clear supply-chain, remote-code-execution/persistence, and data-exfiltration risks.

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

• Third-party content exposure detected (high risk: 0.85). 该技能在运行时会通过 MCP 工具读取抖音/小红书/快手“公开评论”文本（属于第三方平台的外部自由文本），并在后续汇报模板中把 waterfallText、评论原话等内容直接转入 LLM 上下文。

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's MCP configuration instructs the host to run "npx -y ppxc-leads-mcp", which at runtime will fetch and execute remote npm package code (ppxc-leads-mcp), so it is a required external runtime dependency that runs remote code.