Browser Agent — AI for This Page

Browser Agent is a controlled AI assistant for the Chrome Side Panel. It helps you understand page content and complete repetitive browser tasks, while keeping browser actions locked to the page you explicitly attach. What Browser Agent can do: - Read, summarize, extract, and organize page content - Click buttons, fill fields, and navigate within the attached tab - Use selected text and conversation context for follow-up requests - Show execution progress, tool calls, and results - Let you stop tasks and review, export, or delete local history Designed for control: - Each task’s browser actions are locked to the tab where it started - The agent cannot create, switch, close, or control other tabs - The default approval mode asks before sensitive page actions - Purchases, payments, automatic ordering, file uploads, and arbitrary script execution are blocked - Browser-internal and other restricted pages are unavailable Bring your own AI provider: Configure compatible model services using your own API Base URL, API key, and model name. You can manage multiple providers and models and choose which one to use. Optional connections: User-configured MCP read tools and Capability Service workflows can extend supported tasks. You choose the connections and tool permissions; supported structured writes require separate approval. Browser Agent does not provide or operate these optional services. Privacy and storage: Browser Agent does not include advertising or publisher-operated analytics. Configuration, task history, and conversation state are stored locally in your Chrome browser. Relevant task instructions, page content, selected text, conversation context, and connected-service results may be sent directly to your chosen model provider. Connector calls send the required arguments to the service you configure. These providers’ privacy and retention terms apply. Browser Agent helps make browser work faster while keeping page access, approvals, and execution visible to you.