cloudflare-deploy

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

• Potentially malicious external URL detected (high risk: 0.90). The skill demonstrates runtime registration of an MCP server via env.MCP_SERVER_URL (example: "https://mcp.example.com") and calls this.mcp.getAITools(...) to fetch tool definitions which are passed into the AI agent, so external content at that URL can directly control agent prompts/behavior.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The skill explicitly instructs the agent to rerun deployments with escalated sandbox permissions (sandbox_permissions=require_escalated) and offers to request elevated network access, which is a permission-escalation/bypass of sandbox restrictions even though it doesn't ask for sudo or modify system files.