Bot protection for downloads (Turnstile or reCAPTCHA) Pro

A public document library is a scraping target: one script can walk every download link on the page. FileDeck Pro can put a CAPTCHA check in front of downloads — the visitor confirms they’re human once, then downloads normally for an hour.

Two services are supported, and you pick one:

• Cloudflare Turnstile (the default) is free, sets no advertising cookies, and usually verifies without the visitor clicking anything.
• Google reCAPTCHA v2 — the "I’m not a robot" tickbox — if you already run it elsewhere on the site and would rather not add a second vendor.

Whichever you pick covers both this gate and front-end submission: one choice, not two.

FileDeck Pro — Documents → Settings → Downloads → Bot protection.

Setting it up

1. Get keys for the service you want:
   • Turnstile — create a free widget at dash.cloudflare.com → Turnstile.
   • reCAPTCHA — register a v2 ("I’m not a robot") site at google.com/recaptcha.
2. Under Documents → Settings → Submissions, choose the service under CAPTCHA service and paste its site key and secret key.
3. Under Documents → Settings → Downloads, tick Require a quick CAPTCHA check before the first download of a visit. Save.

Until both keys for the selected service are saved, the switch has no effect — the gate does not fail closed on a missing configuration.

Switching service later keeps the keys for the other one, so moving back needs no re-typing. <!– claim-ok: verified by the integration suite — "captcha: switching provider keeps the other set of keys rather than deleting them" –> A value that is neither of the two reads as Turnstile rather than as "no gate". <!– claim-ok: verified by the integration suite — "SECURITY: an unrecognised provider falls back to Turnstile, not to no gate" –>

How it behaves

• The first download click opens a small dialog with the check. Passing it unlocks downloads for one hour; most visitors never see it again.
• Logged-in users are never challenged — they already passed a login.
• Signed share links skip the check, like every FileDeck gate: the link’s signature is its authorisation.
• Multi-select zip downloads are covered by the same check.
• The check is enforced on the server, not just in the browser: a script that calls the download URL directly is refused until it holds a grant.

What it does and doesn’t stop

Be clear-eyed about the strength of this: it is friction, not access control. The grant that a passed check issues is a cookie bound to nothing — not your visitor’s address, not their browser, not a download count — so one person solving one challenge can hand that cookie to a script and download for an hour. What the check buys you is that a scraper must now solve a CAPTCHA to start at all, which defeats the naive ones and costs the determined ones money. If the files themselves matter, restrict them — pair this with access control rather than relying on it.

Within that limit it does stop casual scripted bulk downloading through FileDeck’s download routes and keeps your download counts honest (search-engine crawlers are already excluded from counts, with or without this gate). Like the terms and form gates, it guards the FileDeck download route — a document with no access restriction still has a directly reachable file URL, so pair bot protection with an access restriction when the file itself must not be fetchable.

For metering rather than bot-blocking — "each visitor may download N files a day" — use download quotas; the two work well together.