Docs

Connect Claude to your browser

About five minutes. You need a PagePilot account, a Chromium browser, and Claude Code or a Claude app that supports remote connectors.

1. Add the extension and pair your browser

Create an account (or sign in) on the sign-up page.
Install the extension from the download page and pin it to your toolbar.
Open the extension, choose Cloud (organisation), and select Sign in. It shows a short pairing code and opens the approval page.
Check that the code matches, then approve. Your browser now appears under Dashboard → Devices.
Switch on Let Claude control this browser and allow site access when asked.

2. Add the connector to Claude

Claude Code

claude mcp add --transport http page-pilot https://ai-browser-extension-relay.onrender.com/mcp

Then run /mcp in Claude Code, choose page-pilot, and sign in when your browser opens. Approve access once.

Claude Desktop or claude.ai

Open Settings → Connectors, add a custom connector, and paste this URL. Sign in and approve access when prompted.

https://ai-browser-extension-relay.onrender.com/mcp

Availability of custom connectors depends on your Claude plan. Any MCP client that supports remote servers with OAuth will work.

3. Try it

Ask Claude: “Use the browser tools to open example.com and tell me what is on the page.” Risky actions will open an approval window.

Troubleshooting

“No browser is connected.” Open the extension and check it says Connected. Your browser must be open with control switched on.
“Paused.” Press Resume in the extension popup.
An action was refused. Your organisation’s policy may block that site or tool. Ask an admin.
Several browsers. Claude uses the most recently connected one. Remove devices you no longer use from the dashboard.

Tool reference

Tool	What it does
`browser_list_tabs`	See open tabs (other tabs show only their site).
`browser_select_tab`	Switch to a tab, with your approval the first time.
`browser_new_tab`	Open a page in a new tab.
`browser_close_tab`	Close a tab Claude controls.
`browser_resize_window`	Test mobile and desktop layouts.
`browser_navigate`	Go to a URL (new sites ask first).
`browser_go_back`	Back in history.
`browser_go_forward`	Forward in history.
`browser_reload`	Reload the page.
`browser_snapshot`	A compact, numbered view of text and controls.
`browser_read_text`	The full text of a page or element.
`browser_find_text`	Find and scroll to text anywhere on the page.
`browser_screenshot`	Viewport, element zoom, region zoom or full page.
`browser_wait`	Wait for time, or for text to appear or vanish.
`browser_click`	Left, right, middle and double click.
`browser_click_at`	Click coordinates from a screenshot.
`browser_hover`	Reveal menus and tooltips.
`browser_drag`	Drag and drop by element or coordinates.
`browser_type`	Type into fields (secrets ask first).
`browser_fill_form`	Fill many fields at once.
`browser_select_option`	Choose dropdown options.
`browser_press_key`	Shortcuts like Control+A and Shift+Tab.
`browser_scroll`	Scroll pages, containers, or an element into view.
`browser_upload_file`	Attach files (always asks).
`browser_handle_dialog`	Accept or dismiss alert, confirm and prompt.
`browser_console_messages`	Read console output and errors.
`browser_network_requests`	List the requests a page made.
`browser_evaluate`	Run JavaScript in the page (always asks).
`browser_save_pdf`	Print the page to a PDF.
`browser_cookies`	Get, set or delete cookies (always asks).
`browser_storage`	Read or change localStorage/sessionStorage (always asks).
`browser_set_offline`	Simulate going offline or back online.
`browser_mock_network`	Fake a response for requests matching a URL pattern (always asks).
`browser_status`	Connection, permissions and current tab.