MCP-SHIELD

A research-driven framework to analyze, exploit, and harden MCP servers powering AI agents. Includes vulnerability discovery, adversarial testing, and resilience techniques to secure tool execution, memory, and multi-step agent workflows.

{
  "mcpServers": {
    "mcp-shield": {
      "command": "uvx",
      "args": [
        "mcp-shield"
      ]
    }
  }
}

claude mcp add mcp-shield uvx mcp-shield

{
  "mcpServers": {
    "mcp-shield": {
      "command": "uvx",
      "args": [
        "mcp-shield"
      ]
    }
  }
}