mcp-developer

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

• Third-party content exposure detected (medium risk: 0.30). The MCP protocol workflow described in references/protocol.md ingests JSON-RPC request parameters (e.g., tools/call arguments, resources/read uri) from the connected client over stdio/HTTP/SSE, and those arguments are plain free-form strings/objects that can carry outsider-authored prompt-injection text into the runtime.