A Chrome capture tool with a memory.

Your captures, held on your machine.
Readable by any AI. Owned by none.

catchroll captures full, scrolling web pages cleanly enough that a language model can actually read them — then keeps them in a local library you (or an AI) can come back to. Capture is the easy part. Holding onto what you captured is the point.

Add to Chrome — beta, coming soon How it stays private →

The Web Store listing goes live with the public beta. This page is up first so the link that ships inside every export already resolves.

Capture is the wedge. The library is the moat.

Clean full-page capture

Full-vertical scroll capture with no repeated sticky headers, no UI bleed, no mid-page truncation — dense and legible for an OCR or multimodal model, not just pretty for a human.

An AI-readable library

Captures become portable .catchroll files with the page text alongside the pixels — a personal, searchable record an AI can read directly. Chats end; your library doesn't.

Any AI can read it. No AI owns it.

Export a lean Markdown view and drop it into Claude, ChatGPT, or Gemini. Nothing is locked to one model, one account, or one company's servers.

100% on device.

catchroll keeps your captures in local browser storage and saves your files to your own disk. No remote endpoints. No telemetry. No network calls. Nothing is uploaded, ever — that isn't a setting you turn on, it's how the extension is built.