Deckle’s fetcher

What Deckle is

Deckle is free software for a small e-paper reader, the XTEINK X3, made by one person. Each morning it prints its owner a short paper of five stories from the feeds they chose to follow: your RSS or Atom feed, if they follow it.

Reading your feed is done the usual way, as any feed reader does. This page is about the other thing Deckle does: fetching an article your feed links to, so its owner can read it on their reader.

What it fetches

One copy of an article linked from a feed that someone using Deckle follows, and only one Deckle chose for that person’s paper.
Around 5 AM in that person’s time zone, when their paper is printed.
At most once per article every 3 days, across everyone who uses Deckle: the copy is shared, so a thousand readers of one story cost your site one request. A fetch that fails isn’t tried again for 6 hours.
One request, with no retries. It gives up after 8 seconds, reads at most 3 MB, and accepts only HTML and plain text.
Nothing else on the page: no scripts are run, and no images, stylesheets, cookies or other pages are fetched.

The rules it keeps

robots.txt. Deckle reads your robots.txt first (and keeps it for a day), and obeys the Disallow rules under User-agent: Deckle, or under User-agent: * when there is no group for Deckle. A disallowed article is never fetched: its owner sees the headline only.
No getting around anything. Deckle never tries to get past a paywall, a sign-in or a bot check. It doesn’t change its name, use another address, or look for an archived, cached or AMP copy. A challenge page, a 401, 403, 429 or 503, or an article marked as for subscribers is taken as “no”, and the story is headlines only.
Paywalled sites known to keep everything for subscribers aren’t fetched at all.
It names itself in every request, with the user agent Deckle/1.0 (+https://getdeckle.web.app/bot).

How your words are shown

Only the article’s words are kept: no images, links, ads or layout. They’re shown only on the e-paper reader of someone whose paper carries the story, paged to fit, always with your site named as the source.
The copy is kept for at most 3 days, then deleted. It’s never published on the web, never shown in Deckle’s app, and never used to train an AI model.
In Deckle’s app, the story’s headline links to your article.

Asking to be left out

To keep Deckle from fetching any of your articles, add this to your robots.txt:

User-agent: Deckle
Disallow: /

Deckle sees it within a day. Your stories can still appear in its readers’ papers as headlines from your feed, linking to your site, as in any feed reader. A narrower rule, such as Disallow: /premium/, works too.

For anything else, such as leaving your feed out of Deckle’s catalog altogether, or a problem with how Deckle fetched something, write to the person who runs Deckle. Every message is read.