GetContext

About

GetContext

Turn anything you see on a page into a ready-to-use prompt for your AI coding agent.

AI agents are great at writing code, but blind to your screen. You see the button; they see text. The hard part isn't the AI writing code — it's you transferring the visual context to it. "Change this card" means nothing to an agent. It needs a selector, a file, a route, a component.

GetContext closes that gap. Point at any element on any page, describe the change you want, and it builds a prompt with everything an agent needs to find and edit that element in your source code.


What Makes It Different

Most tools stop at the CSS selector. GetContext goes further:


Ready-to-run ripgrep command — the agent locates the element in your repo instead of guessing.
Framework detection — reads the React fiber / Vue instance to extract the component name and source file.
Captured context — route/URL, accessibility, computed styles, unique selector, XPath.
Optional cropped screenshot of just the element.
Flexible output — Markdown or XML, in three detail levels (Compact / Medium / Extended), with a live token estimate so you see the cost before you paste.



How It Works


Press Cmd/Ctrl + Shift + G.
Hover to highlight, click the element you want.
Describe your change.
Copy the generated prompt and paste it into Claude Code, Cursor, Copilot, or any agent.



Privacy

Uses activeTab, not broad host permissions. It only acts on the active tab when you trigger it. Nothing runs in the background.


Built With

WXT + Preact + TypeScript, Manifest V3, UI isolated in Shadow DOM so it never breaks the page's own CSS.

Free. Built to inspire you to make your own tools — both tools made with AI and tools made for AI.