PenguinHarness Browser

Lets PenguinHarness agents drive the Chrome tabs you hand them: the Penguin tab group and tabs you add with this icon.

PenguinHarness Browser connects your own Chrome to PenguinHarness, so the AI agents you run there can work in real web pages, signed in as you already are. Agents use it through the same "penguin browser" commands they use in the PenguinHarness desktop app's built-in browser: open a page, read it, click, type, and check what changed. It works with a PenguinHarness server you pair it with: the desktop app on your computer, or a server you or your organization run. PenguinHarness is open source: https://github.com/Prism-Shadow/penguin-harness YOU DECIDE WHAT THE AGENT CAN TOUCH • The agent drives only the tabs it opens itself, kept together in a blue tab group named "Penguin", the tabs you hand over with the toolbar button, and the pages those tabs open. It never sees your other tabs or windows. • Take a tab back at any time: drag it out of the Penguin group, close it, or choose "Release this tab" in the popup. "Cancel" on Chrome's debugging bar takes every tab back at once, and "Pause" stops the agent from opening pages or acting in any tab until you resume. • Chrome's own pages, other extensions, the Chrome Web Store and local files are never driven. • The extension itself refuses debugger commands that reach beyond the page: Chrome's cookie store, other targets and the browser itself, storage and request interception, certificate overrides, local file uploads and download settings. A script the agent runs in a page sees only what that page can see. • The debugger attaches when the agent sends its first command to a tab and detaches after 60 seconds without one, so Chrome's debugging bar goes away when the agent is done. PRIVATE BY DESIGN • The extension talks only to the PenguinHarness servers you pair it with. It has no analytics and sends nothing to Prism-Shadow or anyone else. • It injects no content scripts and asks for no access to websites; it reaches a page only through Chrome's debugger, in a tab the agent drives. • What reaches your server is what the agent asks of a driven tab: the page as text or simplified HTML, what its scripts return, screenshots, and the tab's address and title. HOW TO CONNECT 1. In the PenguinHarness Web App, open the Browser panel. It shows the server address and a one-time pairing code, which works once, for 10 minutes. (With a Chrome already paired, choose "Connect your Chrome…" in the panel's menu.) 2. Paste both into the extension's pairing page and choose Connect. The page opens when you install the extension; later, right-click the toolbar icon and choose Options, or use Settings in the popup. You can pair several servers; each one's tabs get a group of their own. The extension's pages are in English, with a switch to Chinese (中文). Requires Chrome 116 or later. Documentation: https://penguin.ooo/docs/builtin-browser