Mcp Watchdog

Mcp Watchdog is a security proxy for Model Context Protocol (MCP) servers that sits between AI coding assistants and MCP servers, protecting against a wide array of known attack classes specific to the MCP ecosystem. It analyzes and filters all JSON-RPC traffic, detecting and blocking threats such as prompt injection, tool poisoning, command injection, SSRF, credential exfiltration, OAuth attacks, and more before they can reach your AI agent. This tool is aimed at developers, security engineers, and organizations using autonomous coding assistants or deploying MCP servers/tools in production environments.