Green claims you can prove.

Reading robot

Who is reading your pages, and how to stop it.

ClaimTraceBot is the automated reader of ClaimTrace. It sends its name with every request, reads public pages at a measured pace to find environmental claims, and can be blocked with two lines in robots.txt. This page says who it is, why it reads and how it behaves.

ClaimTraceBot is the automated reader of ClaimTrace. This page says who it is, why it reads, at what pace, and how to block it in two lines.

Identity

NameClaimTraceBot
Identification sent with every request (User-Agent)ClaimTraceBot/0.1 (+https://www.claimtrace.eu/robot.html)
Operated byAndromède, a French société par actions simplifiée, trading as ClaimTrace (details in the page footer)
Contact (questions about the robot)contact@claimtrace.eu
Page describing the robotThis page

What it reads, and why

ClaimTrace is a European service that checks environmental claims against the rules of Directive (EU) 2024/825. To calibrate and test its rulebook, ClaimTrace measures how environmental claims are really worded on public product pages. For that, the robot reads public pages and keeps a dated copy of the page read (its HTML code), in ClaimTrace's database hosted in the European Union and in internal measurement files, so that a measurement can be justified and reproduced. To find and qualify claims, the text is sent to our model provider, which may process it outside the European Union, under its commercial terms, which exclude the training of its systems on this content. Neither the copies nor any extract are published.

The robot does not look for personal data and reads only public pages; the copy it keeps of a page may however contain some incidentally (a first name in a customer review, for example). It does not log in to any account, does not fill in any form, sends no cookie, runs no script, and does not try to get around a block, a rate limit or a CAPTCHA. The same identification will be used if ClaimTrace one day reads a customer's own pages at that customer's request.

How it behaves

robots.txtRead before any page, once per run. The robot applies the group whose User-agent line carries the name ClaimTraceBot, with or without a version number and whatever the case; otherwise the general group (*). A ClaimTraceBot group that contains only an empty "Disallow:" line allows everything.
If robots.txt is missing or unreadableIf the file does not exist (answer 404 or 410), the robot considers that the site sets no restriction. If it exists but cannot be read (access denied, server error, rate limit, timeout, compressed content, a file larger than 512 KiB, or a web page instead of the file), the robot does not read the site. If robots.txt redirects, the robot follows up to three redirects, including to another site, and applies the rules of the file it arrives at.
PaceAt least 10 seconds between two requests to the same site (www.example.fr and example.fr count as one site), robots.txt included, longer if robots.txt asks for it (Crawl-delay). A site that asks for more than 60 seconds is not read. A site that answers 429 or 503 is left alone for the time given by its Retry-After header, 60 seconds if it gives none, and until the end of the campaign beyond 5 minutes.
VolumeA few dozen different pages per site at most, during a measurement campaign. Campaigns are limited in time. A campaign may re-read the same pages at fixed intervals to measure their changes; it is then announced on this page with its dates, its pace and its ceiling per site.
Current campaignFrom 9 October to 17 November 2026, ClaimTraceBot re-reads 58 public product pages, already read at the end of September, on 40 sites, twice a week, on Tuesdays and Fridays starting at 3:17 UTC: at most 7 pages and 12 requests per site at each run, robots.txt and redirects included (8 when no page redirects), 12 runs in all. These re-reads do not keep the HTML code of the pages: only digests and their useful text, once per version.
Addresses and redirectsPublic addresses only. Three redirects at most; a page is never followed to another site. Size and duration of each download are limited.
Page signalsThe robot ignores meta robots tags and the X-Robots-Tag header: only robots.txt stops or slows it. During the campaign re-reads, it keeps no line of a page that asks for noarchive or none.

How to block it

Add these two lines to the robots.txt file of your site. The robot reads this file before your pages, and stops. The line may carry the name ClaimTraceBot with or without a version number. This is the blocking mechanism the robot supports.

User-agent: ClaimTraceBot
Disallow: /

Going further

To block only part of a site, list the paths with Disallow, encoded as in the address (for example /caf%C3%A9 and not /café). To slow the robot down, add a Crawl-delay line (in seconds, 60 at most) in the group that applies to it.

ClaimTraceBot runs today from the continuous-integration machines of GitHub (GitHub Actions), whose IP addresses change from one run to the next, and, for occasional checks, from a workstation. We do not publish a list of fixed addresses; rely on the User-Agent.

Version 0.1. Last updated: 5 October 2026.

Stay informedOne email at each new stage of the service, never more than one a month.

You will confirm by a link sent by email, and can unsubscribe in one click. The form is loaded from Brevo, our email provider, only when you click.