About Everpage

Most of the web is already gone. Sites redesign, startups die, platforms rot — and the tools that promised to preserve them either saved empty JavaScript shells or locked the copy inside an archive only specialists can open.

Everpage exists to keep the whole thing: every page, forever. Not a screenshot, not a folder of broken links — the complete working site, every file, bootable offline with one double-click. It was built the hard way, against production Three.js worlds and CMS sprawl, until file-level parity and clean headless boots stopped being aspirations and became the test suite.

The name is the promise. If a page was public, Everpage keeps it.

Quickstart

pip install everpage
playwright install chromium   # one-time, for the trace + verify stages

From source:

git clone https://github.com/unknown235669-cmyk/everpage
cd everpage
pip install -r requirements.txt

Interactive menu:

python -m everpage.tui

Download a full website for offline browsing in one command:

python -m everpage.cli https://example.com/ -o ./example-clone --port 8919

Useful flags: --pages about,pricing · --max-pages 200 (sitemap + link crawl cap) · --no-crawl · --no-trace (skip the headless pass — faster, misses runtime-only assets).

Every offline mirror ships OPEN-ME.bat (or python serve_<name>.py <port>) for instant local preview — open the cloned website in your browser with zero setup.

Proven where it matters

Everpage has cloned production sites end to end — Three.js 3D worlds, WebGL portfolios, WordPress portals, Astro CMS sites — and every one boots offline clean. Don't take our word for it: paste a link and watch it happen.

Everpage vs the best website copiers

Researched against the tools' own docs (October 2026): websnap, SingleFile FAQ, Browsertrix docs, independent 2026 roundup.

One independent roundup concluded that a working offline website is delivered by "none of the above." That is the gap Everpage was built to close — and unlike every other row in that roundup, it ships measured proof (see results above) instead of promises.

How it works

python smoke_test.py runs the offline self-checks (no network).

Honest scope

• Backends aren't cloned — APIs are stubbed with captured payloads.
• Gated content isn't reachable — logins, paywalls, DRM and aggressive anti-bot need a real session.
• Live state isn't frozen — websockets, personalization and per-user feeds snapshot to whatever loaded.
• Interaction-gated assets can be missed — the trace pass scrolls and idles; scripted clicking (à la websnap) is on the roadmap.

Layout

everpage/
  cli.py        pipeline orchestration
  fetcher.py    sessions, retries, URL→file mapping, parallel downloads
  detect.py     framework + HTML asset/link extraction
  chunks.py     JS bundle mining (templates, pools, workers, frames)
  rewrite.py    offline path rewriting (+ regression self_test)
  trace.py      headless runtime URL capture
  serve.py      preview server with API stubs
  verify.py     headless boot/error/screenshot check
  tui.py        interactive menu

Contributing

Issues with a failing URL + the probe log tail are gold. PRs that add a self_test regression case alongside any parser change get merged fastest.

License

MIT — see LICENSE.