Browser Automation (Browser Sidecar)

LibrAgent empowers agents to browse the web, search and extract live information, and analyze webpage screenshots autonomously.

---

🛡️ Process Isolation & Stability

Complex web pages or in-page script crashes will never compromise your desktop workspace:

• Isolated Sandbox Execution: Browser automation operates in an isolated background process separate from the main LibrAgent desktop app. Heavy memory consumption or browser crashes cannot freeze or crash your main application.
• Sticky agent profile: browser__createSession with browser="sidecar" (default) reuses a fixed LibrAgent agent browser profile on this device. Logins you make inside that browser survive later sessions until you clear them in Settings → System → Agent browser → Clear agent browser data. This is not your everyday Chrome. Concurrent agent chats share that same cookie jar while the browser sidecar is running.

---

---

✨ New Feature: Everyday Chrome Integration (Stay Logged In)

Instead of opening a blank sandbox browser, agents can directly control tabs in your personal everyday Chrome browser (userChrome).

💡 Key Benefits for Everyday Users

• Preserve Logged-in Sessions: Work directly with services where you are already signed in (Google, GitHub, internal dashboards). You do not need to re-enter passwords.
• Reliable Clicks on Protected Sites: Input simulation bypasses strict Content Security Policies (CSP). Button clicks and keystrokes work reliably on complex websites.
• Save Raw Webpage HTML (saveRawHtml): Saves complete webpage source files to your session workspace for offline inspection and verification.

Execution Mode	Behavior	Best Suited For
sidecar (Default)	Dedicated isolated agent browser	Public web search, documentation research, sandbox tasks
userChrome (Extension Bridge)	Your personal Chrome browser	Tasks requiring personal logins, private intranets, shopping carts

---

🚀 1-Minute Quick Setup

TIP

Chrome Web Store Release Pending (TBD)
The extension is currently under review for the Chrome Web Store. Until the public store link is published, you can install it locally via the 1-minute Load unpacked method below.

1. Locate Extension Path: Open Settings → System → Agent browser and copy the Chrome extension folder path.
2. Install Extension in Chrome:
   • Open chrome://extensions in your Chrome browser.
   • Enable Developer mode in the top right corner.
   • Click Load unpacked and select the extension folder path.
3. Instruct the Agent:
   • Prompt the agent naturally in chat:

     Check the items in my shopping cart in my current Chrome window.

   • The agent switches to userChrome mode and controls your active tab.

NOTE

• After installation, the extension reconnects automatically when you restart LibrAgent.
• Each agent chat maintains exactly one active browser session (Single Source of Truth). Switching between sidecar and userChrome closes previous sessions cleanly.

---

🌐 Key Capabilities

Capability	Description
Page Navigation (browser__navigateToUrl)	Visits specified URLs and detects page readiness to read content.
Screenshot Capture (browser__takeScreenshot)	Captures viewport or full-page images for visual inspection.
Console & Error Tracking	Monitors page JavaScript errors and network state.
Interaction Support (browser__clickElement, browser__inputText)	Supports clicking links, typing text, and scrolling pages.

---

🎯 Practical Use Cases

1. Research Live Documentation:
   • Gathers live release notes, API updates, or competitive news from real web pages.
2. Web UI Verification:
   • Visits local development servers (http://localhost:3000) and captures screenshots to visually verify UI layouts.
3. Automated Daily Briefings:
   • Pairs with Scheduled Tasks to visit specified news sites or dashboards every morning and generate summaries.

---

💡 Notes

• Initial browser launch may take a few seconds while the headless runtime prepares.