OpenClaw Tools (Part 1): Browser Control and Web Search

🌏 中文版

OpenClaw’s agent can control browsers and search the web. This post covers browser management, remote CDP providers, and the Deep Research feature.

Browser Control

Two Profile Types

The managed profile is isolated — it won’t touch your personal browsing data. It’s suitable for most agent use cases.

Configuration

{
  browser: {
    enabled: true,
    profile: "openclaw",       // openclaw | user
    headless: false,           // whether to show the window
    viewport: { width: 1280, height: 800 }
  }
}

Multiple Profile Support

You can configure multiple managed profiles simultaneously for different agents or tasks:

{
  browser: {
    profiles: {
      research: { /* ... */ },
      testing: { /* ... */ }
    }
  }
}

Remote CDP Providers

Don’t want to run Chrome locally? You can use remote CDP (Chrome DevTools Protocol):

Browserless:

{
  browser: {
    provider: "browserless",
    browserless: {
      endpoint: "wss://chrome.browserless.io",
      token: "your-token"
    }
  }
}

Browserbase:

{
  browser: {
    provider: "browserbase",
    browserbase: {
      apiKey: "your-key",
      projectId: "your-project"
    }
  }
}

Sandboxed Browser

The Docker backend supports standalone browser sandbox containers:

• Uses a dedicated Docker network (openclaw-sandbox-browser)
• noVNC observation access is password-protected
• CDP source range can be restricted
• allowHostControl can let sandbox sessions control the host browser

scripts/sandbox-browser-setup.sh

Snapshots and Refs

The browser supports page snapshots (screenshots), which can be referenced by other tools.

CLI Commands

openclaw browser status          # Check browser status
openclaw browser profiles list   # List profiles

Web Search

The agent can use the built-in web search tool to search the internet. Multiple search providers are supported.

Configuration

{
  tools: {
    web: {
      search: {
        provider: "google",    // or other providers
        enabled: true
      }
    }
  }
}

When the provider selection is in auto mode, it checks available API keys in priority order and automatically selects the first usable provider.

Deep Research

A multi-step research mode that combines search and browsing. The agent can:

1. Search for relevant information
2. Browse search result pages
3. Extract key content
4. Synthesize and analyze

This is a skill-level feature, not a single tool — it combines web search + browser + document analysis capabilities.

Summary

Browser and search are the primary ways OpenClaw agents interact with the outside world. Managed profiles ensure isolation, remote CDP avoids local resource consumption, and sandboxed browsers provide security boundaries.

References

This post is compiled from the following OpenClaw source documents:

• docs/tools/browser.md — Browser control
• docs/tools/web-search.md — Web search
• docs/tools/deep-research.md — Deep Research