Web capture

You are reading a paper in your browser. One click puts it in your library: the reference, and the PDF, without a download folder detour or a copy-pasted DOI.

Setting it up#

In Library, at the bottom of the pane, the Web capture row shows whether capture is connected. Click Set up… and follow the two lines for your browser:

• Firefox — Open the add-on shows you the signed file, Copy address gives you about:addons to paste in a new tab, then click the gear and choose Install Add-on From File…. (On macOS nothing is registered to open a .xpi, so Flux reveals the file rather than pretending it can hand it to Firefox.)
• Chrome, Edge or Brave — Show me the folder opens the extension folder, Copy address gives you chrome://extensions to paste in a new tab, then turn on Developer mode and choose Load unpacked.

The status dot turns green the moment your first capture arrives, so you never have to wonder whether it took.

Keeping it up to date#

The extension lives in your browser, not in Flux, so updating Flux does not update it. When capture changes, do it once per browser:

• Firefox — open Set up… → Open the add-on again (or about:addons → the gear → Install Add-on From File…). It upgrades in place; don’t remove the old one first.
• Chrome, Edge or Brave — open chrome://extensions and press Reload on the Flux card.

If you run Flux from a source checkout, npm run build refreshes the folder Chromium loads, and the signed Firefox add-on is in the checkout too — so a git pull plus a build gives you both, and all that’s left is the reload or reinstall above.

Using it#

Open a paper and click the Flux button in your browser’s toolbar. It works on the article page and on a PDF you already have open. The badge tells you what happened:

Supplementary files come with the article in the same click: SI PDFs, spreadsheets, videos. Each is filed against the paper once Flux has identified it — into that paper’s supplements/ folder, which Reader’s Switch PDF menu lists — so a supplement for a paper you don’t have yet simply waits until the paper arrives.

Captures land in a flux/ subfolder of your downloads folder, named flux-…, and Flux removes them as it files them. Flux doesn’t need to be open — anything captured while it’s closed is brought in when you next start it. Nothing is lost, and nothing else in your downloads folder is touched: Flux only ever looks at files it wrote itself.

Flux brings captures in at two moments, both yours: when it starts, and when you press Assign PDFs in Library. That button’s count includes anything still waiting in your downloads folder, so you can always see there’s work to do — and capture while Flux is running as much as you like, knowing your downloads folder won’t rearrange itself under you until you say so.

A captured PDF goes through the same identification as the inbox folder (pdfs_to_assign) — Flux reads the document’s own content (embedded DOI, then the DOI printed on page 1, then its reference list, then a title search) to find the right reference. It attaches to a paper you already have, or adds the reference if it’s new.

Why capture from the browser#

Flux can already fetch open-access PDFs on its own, and paywalled ones through your library proxy. But some publishers sit behind anti-bot systems — Cloudflare and PerimeterX — that refuse automated browsers outright, no matter how the request is dressed up. Those papers are simply unreachable to Flux.

Your browser is not an automated browser. It’s you: already signed in, already past the paywall, already holding the session. The extension captures the PDF from inside your own session, which is why it reaches papers Flux cannot fetch by itself. It’s the most capable acquisition route in the app, and it costs one click.

Browser support#

If nothing happens#

• Hover the toolbar button. The tooltip carries the reason the last capture failed, including the browser’s own error code when a publisher refused the download.
• Check the address bar for a blocked-download icon, and allow it.
• Some publishers refuse the PDF but not the page. You get the amber badge and the reference still arrives — just without its PDF. Flux can often fetch that separately with Get PDFs.
• Internal pages can’t be captured (about:, chrome://, the extensions page itself) — browsers don’t allow it.

Where things land#

Nothing is ever guessed: a PDF Flux can’t identify with confidence is set aside rather than attached to the wrong paper.