Publication reading
Last updated: October 16, 2026.
Desvelio reads the public side of news publications. This page says what is read, what is kept, and how a publication switches on the legal refusal.
Who
Desvelio is a tool for newsrooms. The operator is named in the legal notice. The reading is done by our own server, to see what the rest of the market is publishing.
What is read
- The sitemap, the public feeds, the front page and the section fronts.
- The most-read box, when the page shows one.
- Each article, once.
- robots.txt, ads.txt and sellers.json.
What is kept
- The address, the headline, the times, the section, the position, the metadata, the counts and the labels.
- A raw copy of fronts and feeds, for seven days. Never of articles.
- The article text is not stored, not sent to anyone else, and not republished. A headline is kept with a link to the original.
- Nothing behind a login or a paywall is opened. Advertising and measurement on the page are not loaded.
How often
The front page every 15, 30 or 60 minutes, depending on the publication’s size. Each article once. Up to ten requests a second to one site, four at a time, and a safety ceiling of 20,000 requests a day so a broken list cannot knock a site over.
From where
From our own server in the European Union. Pages are opened the way a phone opens them. The address is not printed here.
How to stop it
The legal refusal is a robots.txt rule for everyone, or a text-mining reservation. This block stops the reading on the next pass:
- User-agent: *
- Disallow: /
A text-mining reservation (the file /.well-known/tdmrep.json, or the tdm-reservation header) keeps the counts and drops the headlines. A request that names the publication deletes the headlines we stored for it. The contact form is how to send that request. How we treat people who write to us is in the privacy policy.