@radware/openclaw-guard

OpenClaw plugin for Radware Agentic Protection — detects prompt injection attacks in real time.

Installation

openclaw plugins install @radware/openclaw-guard

Configuration

Add the plugin config to ~/.openclaw/openclaw.json:

{
  "plugins": {
    "entries": {
      "radware-guard": {
        "enabled": true,
        "config": {
          "apiKey": "sk-rdwr-...",
          "apiUrl": "https://..."
        }
      }
    }
  }
}

You can also configure the plugin through the OpenClaw Control UI at http://127.0.0.1:18789 (Config tab).

Required fields

| Field | Description | |---|--------------------------------------------| | apiKey | Your Radware API key (provided by Radware) | | apiUrl | Radware API endpoint (provided by Radware) |

Optional fields

| Field | Default | Description | |---|---|---| | timeoutMs | 30000 | Request timeout in milliseconds | | mode | monitor | Plugin mode — currently only monitor is supported |

Restart the gateway

After configuring, restart the gateway:

openclaw gateway --force

How it works

The plugin hooks into OpenClaw's tool-call lifecycle and sends each pending tool call to the Radware Agentic Protection API for inspection. Radware evaluates whether the tool call was driven by the user's legitimate intent or by injected instructions hidden in external data (web pages, file contents, API responses).

Verifying it works

Trigger any tool call and look for log output:

radware-guard: outgoing request { "UserPrompt": "...", "ToolName": "...", ... }

License

UNLICENSED