tabshot

Lets a local CLI photograph allow-listed tabs and click, type, scroll in them. Never hands over page text, cookies or URLs.

tabshot lets a program on your own computer — a script, or a coding assistant running in your terminal — look at one of your browser tabs and act on it, without ever handing it the page itself. Pixels in, coordinates out. The command line can receive: • a PNG of the tab's visible area (or a zoomed crop of it) • "ok" or an error • whether a tab for a domain is open It cannot receive page text, the page's HTML, cookies, form values, URLs, your list of tabs, or anything from a domain you have not allowed. There is no "evaluate" and no "get text" — the protocol has no way to ask. What it can do in an allowed tab: click at a point, type text, press keys, scroll, refresh, resize the viewport, and hand local files to a file field. Your sign-ins stay in your browser, so it works on admin panels and dashboards you are already logged in to. How it works • You run the free, open-source tabshot command-line program (github.com/faradey/tabshot). It listens on 127.0.0.1 only and creates a token. • Paste the token into the extension's options and list the domains it may touch. Chrome asks you to grant access to exactly those domains. • Click the tabshot icon on a tab to share it. Commands run in a separate, unfocused window tabshot opens itself — your own tab is never moved. Every command is recorded in the extension with a picture of the tab after it, so you can watch back what was done or download it as a video. Recordings never leave your browser; text typed into password and card fields is kept as its length only. Requires the tabshot command-line program on the same computer. Without it the extension does nothing.