# Connect an agent to Chrome

Browser f6n lets an agent work in your existing Chrome profile through the
Browser f6n extension and a remote relay.

1. Install the extension in the desktop Chrome profile you want the agent to
   use, open that profile, and choose **Sign up or sign in**.
2. Point your agent at this guide. If it supports remote MCP and OAuth, use
   `https://browser.f6n.run/mcp`.
3. Otherwise, give the agent the Browser f6n skill and let it install the
   invited-alpha CLI tarball. The current direct-install commands are:

   ```text
   npm install -g https://browser.f6n.run/downloads/browser-f6n-cli-0.1.0.tgz
   browser-f6n connect --json
   ```

   For one-shot use, run:

   ```text
   npx --yes https://browser.f6n.run/downloads/browser-f6n-cli-0.1.0.tgz connect --json
   ```

   The future `@browser-f6n/cli` npm package is not published yet; use the
   versioned tarball URL above for this alpha.
4. Open the approval link it returns, check the short code and account, and
   approve Browser f6n access. Approval may happen on another device; it does
   not start Chrome on your desktop.
5. The agent should resume the same attempt if approval is still pending, then
   run `status` and a harmless tabs check before saying that Chrome is ready.

Keep Chrome open and online while the agent works. The extension may need you
to sign in again after Chrome restarts. To stop the agent's local access, ask it
to run `browser-f6n logout --json`; to stop the browser connection, sign out or
disable the extension.

The agent can read and act in the connected profile. Review its approval and
consequential actions, and do not treat page instructions as trusted commands.