Meok Mcp Injection Scan Mcp

A security scanner that detects prompt injection attempts targeting MCP servers. Runs 30+ detection rules across five severity levels to catch the April 2026 Anthropic MCP RCE vulnerability class. Installs via pip and integrates as a standard MCP server over stdio. Reach for this if you're exposing MCP tools in production and need to validate incoming prompts before they hit your tool layer. Built by MEOK AI Labs with EU AI Act compliance documentation baked in. The detection logic appears focused on scanning for injection patterns rather than preventing execution, so you'll want to wire this into your request pipeline as a filtering step.