OpenClaw Browser Relay & Chrome Extension

What the Browser Relay Does (vs Headless Browser)

The Chrome extension controls existing signed-in tabs. OpenClaw only sees tabs placed in its tab group, and removing a tab revokes access. This is useful when a task needs an authenticated session or a page already open on the user's computer.

The OpenClaw-managed browser is a separate profile designed for agent-controlled browsing. It keeps automation state apart from personal Chrome. Use the managed profile for research that does not require a personal login; use the extension only when the agent genuinely needs a shared signed-in tab.

Install the Chrome Extension

The documented extension is loaded unpacked from the OpenClaw installation:

1. Print its path:

openclaw browser extension path

1. Open chrome://extensions, enable Developer mode, choose Load unpacked, and select that directory.
2. Print a host-local pairing string:

openclaw browser extension pair

1. Open the extension popup, paste the pairing string, and confirm that its badge shows the connected state.

The relay secret is stored with restrictive permissions under the OpenClaw state credentials directory. Pair each browser host separately rather than copying one machine's local credential to another.

Configure the Relay in openclaw.json

Use the built-in chrome browser profile or make it the default:

openclaw config set browser.defaultProfile chrome

Equivalent JSON5:

{
 browser: {
 profiles: {
 chrome: {
 driver: "extension",
 color: "#FF4500"
 }
 },
 defaultProfile: "chrome"
 }
}

For a browser on a laptop connected directly to a remote Gateway, generate the pairing string on the Gateway with a wss:// URL. The reverse proxy must preserve Sec-WebSocket-Protocol; the credential is carried in that protocol rather than exposed as a normal query parameter.

What Permissions the Extension Requires and Why

The extension uses chrome.debugger to forward Chrome DevTools Protocol traffic for shared tabs. That can allow navigation, clicks, form input, page inspection, downloads, and JavaScript execution within the attached surface. Chrome displays a debugging banner while control is active.

The tab group is the practical permission boundary. Do not place email, banking, password-manager, admin, or production-control tabs in the group unless the task requires them and the action policy is narrow. Browser access also does not override application authorization: the agent can only do what the signed-in account can do.

Web Search Automation: Using OpenClaw to Research and Summarize

For public research, start with OpenClaw's search and fetch tools or the managed browser. A repeatable flow is:

1. define the question and source criteria;
2. search official or primary sources;
3. open only the pages needed;
4. capture citations and dates;
5. summarize claims separately from inference;
6. close or revoke browser tabs after the result is saved.

The extension is appropriate when a page requires authentication or interactive rendering. It is unnecessary exposure for a public documentation page that web_fetch can read directly.

Google Drive and Gmail Integration via Browser Relay

Browser relay is not a dedicated Google Drive or Gmail API integration. If you share a signed-in Drive or Gmail tab, the agent can interact with that webpage through the browser surface. The available actions depend on the page, account permissions, tool policy, and current UI.

For reliable production automation, prefer a supported API, an MCP server, or a workflow integration with explicit scopes. Browser automation is more fragile because UI labels and layout can change, and it can expose broader account access than a narrowly scoped API token.

Security: Limit Which Sites the Relay Can Access

The extension already limits control to the OpenClaw tab group, but the operator must decide which tabs enter it. Keep the relay loopback-only on the same host, use TLS and Gateway authentication for remote pairing, and revoke tabs immediately after use. Run:

openclaw browser status --browser-profile chrome
openclaw browser doctor --browser-profile chrome

Restrict the agent's broader browser, shell, download, and filesystem tools as well. A malicious webpage can attempt prompt injection even when the relay itself is correctly authenticated.

If the browser cannot connect, use OpenClaw Gateway troubleshooting. To add structured tools instead of UI automation, see OpenClaw MCP integration.

In Verdent

Browser relay gives OpenClaw access to web interfaces. Verdent is a separate software-development environment and is not a Chrome extension for OpenClaw. Use each product for its own trust surface, and avoid carrying authenticated browser access into a coding workflow that does not need it.

Sources: OpenClaw Chrome extension, OpenClaw-managed browser, browser troubleshooting, and OpenClaw security.