mcp-builder

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

• Third-party content exposure detected (high risk: 0.80). The skill instructs the agent to fetch and load public web documentation (e.g., https://modelcontextprotocol.io/sitemap.xml and raw GitHub files at https://raw.githubusercontent.com/...) and to use WebFetch/web search to load external pages, so the agent will read untrusted public web content that could carry indirect prompt injection.