MCP Integration

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly directs connecting to external MCP servers via SSE/HTTP/WS (e.g., https://mcp.asana.com/sse, https://api.example.com/mcp, wss://mcp.example.com/ws) and describes agents that autonomously call and interpret those external MCP tools/results (see "Using MCP Tools in Commands and Agents" and the "Agent: data-analyzer.md" workflow), so untrusted third‑party content can be fetched and materially influence agent decisions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill configures runtime MCP server endpoints (e.g., https://api.example.com/mcp) which Claude Code connects to to fetch tool schemas and invoke remote tools that directly determine agent behavior (and the stdio examples such as using npx show remote code can be executed at runtime), so this external URL is a runtime dependency that can control prompts/execution.