MCP Defender

MCP Defender is a free, open-source (AGPL-3.0) security proxy sitting between AI applications and MCP servers, analyzing communications in real time for threats like prompt injection, credential theft, and command injection — recently acquired by Docker Inc. Who it's for: developers using MCP (Model Context Protocol) servers with AI tools like Cursor, Claude, VS Code, or Windsurf who want real-time threat monitoring on that communication layer, a genuinely new and still-maturing attack surface. Strengths: being fully open source (AGPL-3.0, code on GitHub) means the actual detection logic is auditable rather than a black box, which matters for a security tool specifically. The recent acquisition by Docker is a real, concrete signal of continued investment and backing rather than an abandoned side project. Weaknesses and limitations: MCP itself is a relatively new protocol, so the threat landscape it's designed to catch is still evolving — this addresses known attack patterns as of now, not a guarantee against future novel ones. Real-world scenario: a developer using MCP servers with an AI coding assistant wants real-time protection against a malicious or compromised MCP server attempting prompt injection or credential theft.