Week 2026-W31
2026-07-27 00:00:00 UTC to 2026-08-02 23:59:59 UTC — the week is still in progress. Every figure on this page is computed from the record when the page is requested, not written down afterwards.
The week is still in progress, so this is a partial count against a full one (572 last week). Nothing can be read from the direction yet.
The largest single address contributed 893 of those requests — 37.4%. Anyone can add to this count; nothing here refuses a client, because refusing traffic would change what the instrument can measure. So the concentration is published beside the total rather than prevented.
Declared identities
| Declared an AI crawler identity | Requests |
|---|---|
| Corroborated by the vendor's published range | 67 |
| A different range belonging to the same vendor | 0 |
| Contradicted — vendor publishes a list, this address is not on it | 0 |
| Uncheckable — no published list exists | 77 |
| Total | 144 |
Checked against the vendor snapshot captured 2026-07-26, never a live fetch, so this table reproduces. Only contradicted is evidence against a client, and uncheckable is a gap in a vendor's publishing rather than anything about the client. The running totals carry the same split.
Behaviour
| Measurement | This week |
|---|---|
| Fetches of paths disallowed in robots.txt | 1 |
| Requests carrying a conditional header | 1 |
Both figures are small and both are honest about why. The disallowed paths serve ordinary content and are listed so that compliance is measurable rather than assumed. Conditional requests can only be counted on the formats where the validator survives the CDN — which is not all of them.
Findings decided
15 published:
One user agent, 29 addresses, 20 paths — a retrieval spread thin enough to look like nothing
A single user agent string arrived from 29 distinct addresses in 2 countries over 44 hours. 28 of those addresses sent exactly one request. Between them they fetched 20 distinct paths.
One address presented crawler identities belonging to 10 different companies
A single connecting address made 90 requests under 13 different crawler user agents, belonging to 10 separate companies, across 87 distinct paths. Crawlers operated by different companies do not share an address.
One address requested 34 distinct paths within 3 seconds
A single connecting address took 34 distinct paths in 3 seconds, a rate no interactive session produces.
The most thorough reader of this site declares itself a 2019 handset, from ten countries at once
One user agent string — a consumer iPhone running iOS 13.2.3, released November 2019 — has made 222 requests to this site from 168 distinct addresses in ten countries, averaging 1.32 requests per address. It has taken 49 distinct paths, read robots.txt, taken no disallowed path, executed no script and never asked whether anything had changed. It is the only client that has fetched all three pages published here on 27 July. A single handset is one device in one place; the shape is the part that cannot be reconciled with the declaration.
Fourteen sites served the same 501 bytes of robots.txt, and the bytes name who wrote them
A declared sample of 400 domains was asked for its robots.txt on 28 July 2026. 198 answered with one. Fourteen of those files contain 1,834 bytes that are identical across all fourteen, at the same offset — but only 501 of them sit inside the comment marking what the delivery network inserted. The marked part refuses eight AI crawlers. The 1,333 unmarked bytes before it are a legal notice, written in the site's own voice, asserting terms as a condition of access and reserving rights under EU copyright law. In none of the fourteen does the text outside the marked block name any of those crawlers, so nothing was overruled: the decision appears in a file the owner publishes and is absent from everything the owner wrote.
Eight requests from OpenAI's two automated crawlers asked only for the rules and the map
Between 25 and 27 July 2026, requests declaring OAI-SearchBot arrived here six times from five addresses on three separate days, and every one of them asked for /robots.txt. Requests declaring GPTBot arrived twice and both asked for /sitemap.xml. Neither crawler has fetched a single page of this site's content. The only OpenAI identity that did is ChatGPT-User, which fetches when a person asks it something. A further eleven requests declaring OAI-SearchBot are excluded: they came from one address that presented thirteen different companies' crawler identities inside a minute.
Every request declaring an OpenAI identity was answered 200, and none has declared the browsing agent since 25 July
Eleven requests declaring one of three OpenAI identities are in this record, each corroborated against a dated snapshot of that vendor's published addresses, and every one was answered 200. None was refused, redirected or rate-limited. The CDN's event log for the last 24 hours contains a single block, of an address that appears nowhere in this record and is not on any OpenAI list. Requests declaring the browsing agent stop on 25 July and have not resumed, while an assistant asked repeatedly to open this address reported that it could not. Nothing measurable from this side accounts for that.
893 requests in three minutes became 77% of a day's traffic, and obtained four files anyone can read
Between 11:39:38 and 11:42:49 UTC on 28 July 2026 a single address sent 893 requests asking for 884 distinct paths. 889 were answered 404. The four that were not returned the home page, robots.txt, llms.txt and the sitemap — the files this site publishes for anyone. In the whole record, no request for a credential, configuration or administrative path has ever been answered with anything below 400, no request has ever produced a server error, and all 24 requests using a writing method were answered 404. The finding is not the scan. It is that those three minutes were 77% of the surrounding day, on a site whose published figures are counts of requests.
One user agent, 13 addresses, 13 paths — a retrieval spread thin enough to look like nothing
A single user agent string arrived from 13 distinct addresses in 1 country over 21 hours. 13 of those addresses sent exactly one request. Between them they fetched 13 distinct paths.
One address requested 84 distinct paths within 23 seconds
A single connecting address took 84 distinct paths in 23 seconds, a rate no interactive session produces.
One address requested 22 distinct paths within 3 seconds
A single connecting address took 22 distinct paths in 3 seconds, a rate no interactive session produces.
One address requested 47 distinct paths within 5 seconds
A single connecting address took 47 distinct paths in 5 seconds, a rate no interactive session produces.
Which content formats are actually fetched
Across 43 fetches of identical content published in several formats, probe_llms_txt was requested most often (13).
One address requested 31 distinct paths within 6 seconds
A single connecting address took 31 distinct paths in 6 seconds, a rate no interactive session produces.
One address requested 165 distinct paths within 17 seconds
A single connecting address took 165 distinct paths in 17 seconds, a rate no interactive session produces.
2 withdrawn or rejected. The subject and the reason for each are listed on the findings page.
Rejections are counted here and never named, because publishing an unverified statement about a client in order to prove it was rejected repeats the offence. The count is the part that matters: a review step that has never rejected anything has not been shown to be one.
Markers
25 new markers published this week, 72 live in total. 0 have ever been observed in a language model's output.
That last number is the one this site exists to move, and it has been zero since the first day. It is printed every week whether or not it changes, because a counter that only appears when it is interesting is a counter nobody can trust. What a marker is.
Other weeks
- 2026-W31 — this week
- 2026-W30
Each week keeps its own address permanently, so a figure quoted from here can be checked against the week it was quoted from. Reports are recomputed on every request, which means an older week's page may change if the record behind it is corrected — the figures are a view of the record, never a snapshot taken away from it.