@moltnet/guard

MoltGuard plugin for Clawdbot — Security & observability for AI agents.

Features

• 🔍 Action Logging — Every tool call is logged with risk classification
• 🧠 Mind Graph — Visualize your agent's thought process in real-time
• ✋ Intent Gating — Require approval for high-risk actions
• 🎮 Remote Control — Pause, resume, or stop your agent remotely
• 📊 Audit Dashboard — Complete history with search, filters, and export

Installation

clawdbot plugins install @moltnet/guard

Configuration

1. Get your token at guard.moltnet.ai

2. Add to your Clawdbot config (~/.clawdbot/config.json):

{
  "plugins": {
    "entries": {
      "guard": {
        "enabled": true,
        "config": {
          "token": "mg_your_token_here",
          "agentName": "my-assistant"
        }
      }
    }
  }
}

3. Restart Clawdbot

clawdbot gateway restart

Configuration Options

| Option | Type | Default | Description | |--------|------|---------|-------------| | token | string | required | Your MoltGuard API token | | url | string | https://guard.moltnet.ai | MoltGuard server URL | | agentName | string | clawdbot-<hostname> | Display name in dashboard | | logThoughts | boolean | true | Send reasoning traces to Mind Graph | | gateHighRisk | boolean | true | Require approval for high-risk actions | | gateTools | string[] | ["exec", "message", "write"] | Tools that require approval | | pollCommands | boolean | true | Enable remote control (pause/resume/stop) | | pollIntervalMs | number | 5000 | Command poll interval in ms |

Risk Classification

The plugin automatically classifies tool calls by risk:

| Risk | Tools | |------|-------| | 🔴 Critical | gateway, exec (with rm, sudo) | | 🟠 High | exec, message, nodes | | 🟡 Medium | write, edit, browser, cron, process | | 🟢 Low | read, web_search, web_fetch, memory_* |

How It Works

┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│    Clawdbot     │────▶│   MoltGuard     │◀────│     Human       │
│    (Agent)      │     │   Plugin        │     │   (Dashboard)   │
└─────────────────┘     └────────┬────────┘     └─────────────────┘
                                 │
                                 ▼
                        ┌─────────────────┐
                        │  guard.moltnet  │
                        │     .ai         │
                        └─────────────────┘

1. Plugin intercepts all tool calls via tool_result_persist hook
2. Actions are logged to MoltGuard with risk classification
3. High-risk actions wait for approval (if gateHighRisk enabled)
4. Human approves/rejects from dashboard or Telegram
5. Plugin polls for remote control commands

Self-Hosting

If you're running your own MoltGuard instance:

{
  "plugins": {
    "entries": {
      "guard": {
        "enabled": true,
        "config": {
          "url": "https://your-moltguard.com",
          "token": "your_token"
        }
      }
    }
  }
}

Links

• Dashboard: guard.moltnet.ai
• Docs: guard.moltnet.ai/docs
• GitHub: github.com/moltnet/guard

License

MIT