agents-get-started

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

• Third-party content exposure detected (high risk: 0.80). The skill's scaffold (Step 4, "Heads-up on the scaffolded MCP client") explicitly includes a pre-wired mcp_client that points at a public example MCP endpoint, meaning the agent will call and ingest responses from an external public endpoint that could contain untrusted or user-generated content capable of influencing tool behavior and decisions.