Documentation Menu
Browser Overview
Watch your agent browse the web in real time.
On this page
Your BrainRoad agent has a built-in headless browser it uses for web tasks. The Browser tab in the dashboard lets you see exactly what the agent sees as it navigates the web.
What the Browser Tab Shows
The Browser tab streams a live view of the agent’s headless browser session directly to your dashboard. When the agent opens a webpage, fills out a form, clicks a link, or reads content, you see it happen in real time.
This is not a screenshot or a delayed summary. It is a continuous visual feed of the agent’s browser activity.
When the Browser Is Useful
The browser view is valuable when your agent is:
- Researching topics — Watch it navigate search results, read articles, and gather information across multiple sites.
- Filling out forms — See exactly which fields the agent fills in and what values it enters.
- Navigating web applications — Observe the agent interact with dashboards, admin panels, or online tools.
- Debugging web tasks — If the agent gets stuck on a page or encounters an error, you can see exactly where things went wrong.
How It Works
Your agent runs a headless Chromium browser in a sandboxed sidecar container. When you open the Browser tab, the browser’s visual output is streamed to the dashboard over a WebSocket connection. The feed updates continuously while the agent is actively using the browser.
When the agent is not performing any web tasks, the browser view shows an idle or blank state. The browser only activates when the agent decides it needs to visit a webpage as part of completing a task.
Chrome DevTools Protocol
For developers who want programmatic browser control, the browser sidecar exposes the Chrome DevTools Protocol (CDP) on port 9222 inside the agent’s container. You can connect to it from the Console to automate browser interactions, capture screenshots, or run JavaScript in the page context.
# From the Console, check the browser endpoint
curl http://localhost:9222/json/version
This is useful for building custom automation scripts, running Puppeteer/Playwright inside your agent’s environment, or debugging browser behavior programmatically.
Related Docs
- Using the Browser Tab — how to access and interpret the browser view
- How Agents Work — understand the agent’s isolated environment including its browser
- Activity Overview — monitor your agent’s actions alongside the browser view