About CopyAnySite

Why it exists

Every tool that claimed to copy a website was really doing one of two things: downloading the HTML a server sends, or archiving a page as a bundle of files. Neither survives the test that matters — open the copy with the original site unreachable and see whether it still looks like the original. Modern pages assemble themselves in the browser, so a copy taken before that assembly is a copy of nothing.

CopyAnySite starts from the other end. It captures the page as your browser has already rendered it: the live DOM, the computed styles that apply to it, the fonts and sprites and background images and posters it actually paints. Then it rebuilds that into something you can keep — one HTML file, a ZIP, an editable WordPress page, or a kit a coding agent can work from.

How we measure, and why we publish the losses

Every fidelity number on this site comes from npm run bench:rivals in our own repository: eight real pages, three runs, four tools (CopyAnySite, SingleFile, monolith and GNU wget), each copy rendered with the original origin blocked at the network level and scored by pixel comparison against the source.

On that benchmark CopyAnySite averages 98.87%, against 87.93% for SingleFile, 87.40% for wget and 83.44% for monolith. On simple pages the field converges — Hacker News is 100% for three of the four tools, and nobody deserves credit for winning a page with no webfonts and no JavaScript. The distance opens up on animation-heavy marketing sites and commercial WordPress themes, which is where the honest pitch lives.

When a competitor wins a row, the row stays on the page. A benchmark you can only lose is marketing, not measurement — and ours is in the repository so you can run it yourself.

What we will not build

No paywall bypass, no authentication bypass, no DRM stripping, no scraping of personal data. The extension captures what your own browser is already showing you, and refuses hosts on an operator block list. That boundary is not a legal hedge, it is the product: a tool that crossed it would be unusable by the professionals it is built for.

We also do not store what you capture. Capture and export happen in your browser; the control plane sees licence checks and, if you leave telemetry on, hostnames and counts. Never URLs, never content.

Who is behind it

Karandeep Singh is the founder. The company operates from New York, NY 10080, USA, and the source for the capture engine, benchmark and control plane lives at github.com/bytepassperks/sitegrab.

Contact

Questions, bugs, licensing, press: use the contact form. It reaches the support inbox directly, and a bug report with a URL in it is the fastest way to change the product.