PageScrape

Paste one URL. Get a spreadsheet.

PageScrape walks every page of a listing, auto-detects the columns, and exports one clean table to Excel. No selectors, no per-row credits, no account — the default run is fully local.

Add to Chrome Privacy policy

Manifest V3 · auto-detects records with structural heuristics, not an LLM · your data stays on your device

Paste one URL

PageScrape reads the first page, auto-detects the repeating records and their fields, and shows you a preview table.

No selectors, no code

Detection uses proven structural heuristics (MDR/DEPTA), not an LLM — so it just works on tables, lists, cards, and messy registry markup.

Edit the columns

Rename, drop, or reorder columns — or re-pick the record region if the first guess wasn't quite right.

Automatic pagination

It detects the scheme — ?page=N, a JSON endpoint behind infinite scroll, or a real "Next" button — and shows it for confirmation before it runs.

Resumable crawls

A crawl survives a browser restart or a service-worker nap. Reopen the panel and click Resume from the last saved page.

Export anywhere

Send your table to .xlsx, .csv, or the clipboard. Rows persist on-device, so a failed export never loses a run.

Optional AI cleanup — only when you ask

If a column is messy, bring your own Claude, OpenAI, or Gemini key to relabel or tidy it. Only the columns you pick are sent, using your own key, and a live cost meter shows every token and dollar. It's off by default and never fires unless you turn it on.

Private by design

No account. No PageScrape server, no analytics, no tracking. PageScrape fetches only the pages you point it at, from your own browser, and your scraped tables are stored only on your device. Data leaves your browser only when you export a file. Read the privacy policy →