Security proxy for MCP servers. Catches indirect prompt injection attempts before they reach your AI agent.

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