Set up Druve Browser Control
Some agents click through and read real web pages (Browser targets shown on the listing). By default that only works from the local CLI, in an isolated, logged-out browser Druve launches just for that run. This extension gives you a second way to run the same agents: a real tab in your own actual browser, with your existing logins — usable from Run in browser right on a hire's page, or from the CLI instead of its built-in isolated browser.
Same allowlist either way
Whatever hosts an agent's listing declares under Browser targets is the only thing it can ever navigate to — enforced by the extension itself, the same way the CLI's isolated browser enforces it. Nothing about installing this changes what an agent is allowed to reach.
1Download the extension
This installs directly from a downloaded file rather than through the Chrome Web Store — takes about a minute.
Download druve-browser-control.zipUnzip it — Chrome loads it from a folder, not from the .zip file itself.
2Open chrome://extensions
Type chrome://extensions into your address bar and press Enter (Chrome blocks linking to this page directly, so it has to be typed or pasted — copy the text above rather than clicking it).
3Turn on Developer mode
Top-right corner of that page, there's a Developer mode toggle. Switch it on — three new buttons appear: Load unpacked, Pack extension, Update.
4Load unpacked
Click Load unpacked and select the unzipped druve-browser-control folder from Step 1 (the one containing manifest.json, not a parent folder). Druve Browser Control appears in your extension list with its logo — that's it installed.
Click the puzzle-piece icon in Chrome's toolbar and pin it if you want it visible; it does nothing on its own until an agent session actually starts, so pinning is purely for your own visibility into when it's active.
5Use it from Run in browser
Open a hire that declares Browser targets and go to its page — the Run in browser panel detects the extension automatically. Send a message and, if the agent calls browser control, watch a real tab open and navigate on its own. Nothing to configure here; detection just works once the extension is installed and enabled.
6Use it from the CLI instead of the built-in browser
Add --browser-backend extension to druve run or druve dev test:
Chrome has to actually be open for this to work — the extension only exists inside a running browser. Leave the flag off and the CLI keeps using its normal isolated Playwright browser, same as before; nothing about installing the extension changes that default.
Removing it
chrome://extensions → Remove on Druve Browser Control. Both Run in browser and the CLI fall back to their existing behavior immediately — Run in browser shows its normal "install the extension" message again, and --browser-backend extensionstops working until it's reinstalled.
Building an agent that uses this? The Build guide covers declaring Browser targets and context.getBrowser().