CrawlCheck
Data · census · weekly

Agentic web census, week 2026-W47

Week to date; sealed after Sunday, midnight UTC. The same frozen sets of the most-visited sites, measured on a schedule from outside: what they tell and serve AI agents this week against last week. No site is named.

top 1,000, daily

Run 52 (finished 2026-11-22), 991 sites, compared with run 45 (2026-11-15).

Biggest moves: At least one AI identity is refused (non-2xx) where the browser got 2xx -1 pt (2 gained, 7 lost); Serves /llms.txt +0.8 pt (4 gained, 0 lost); GPTBot refused where the browser got 2xx -0.6 pt (2 gained, 5 lost); OAI-SearchBot refused where the browser got 2xx -0.6 pt (2 gained, 5 lost); robots.txt names the AI-training policy token (__usage_ai-training, IETF AIPREF draft) +0.5 pt (2 gained, 0 lost).

FigureThis weekChange, same sitesSites +/−
Homepage answers CrawlCheck's census fetch (2xx after redirects)
of 991
52.3%0 pt+11 / −11
Homepage refuses CrawlCheck's census fetch (400, 401, 403 or 429): blocked or rate-limited, not down
of 991
12.2%+0.3 pt+7 / −4
No website answers at that name (5xx, 530, timeout or no answer)
of 991
31.6%-0.2 pt+1 / −3
Serves a robots.txt (200, not HTML)
of 518
78.8%-0.2 pt+1 / −2
robots.txt names a sitemap
of 408
71.6%0 pt+1 / −1
robots.txt carries a Content-Signal line
of 408
4.7%+0.3 pt+1 / −0
Content-Signal says ai-train=no
of 408
2.9%0 pt+0 / −0
Content-Signal says ai-input=no or search=no
of 408
1%0 pt+0 / −0
robots.txt names the AI-training policy token (__usage_ai-training, IETF AIPREF draft)
of 408
2%+0.5 pt+2 / −0
robots.txt states a preference against AI training on / with that token
of 408
1.2%0 pt+0 / −0
robots.txt closes / to at least one of the 12 AI agents
of 408
25%+0.3 pt+1 / −0
robots.txt closes / to GPTBot
of 408
16.2%0 pt+0 / −0
robots.txt closes / to OAI-SearchBot
of 408
7.8%0 pt+0 / −0
robots.txt closes / to ChatGPT-User
of 408
9.6%0 pt+0 / −0
robots.txt closes / to ClaudeBot
of 408
15.4%0 pt+0 / −0
robots.txt closes / to Claude-SearchBot
of 408
9.3%0 pt+0 / −0
robots.txt closes / to PerplexityBot
of 408
12%0 pt+0 / −0
robots.txt closes / to Google-Extended
of 408
14%0 pt+0 / −0
robots.txt closes / to Applebot-Extended
of 408
12.5%0 pt+0 / −0
robots.txt closes / to CCBot
of 408
19.1%0 pt+0 / −0
robots.txt closes / to Bytespider
of 408
19.1%+0.3 pt+1 / −0
robots.txt closes / to Meta-ExternalAgent
of 408
15.2%+0.3 pt+1 / −0
robots.txt closes / to Amazonbot
of 408
13.2%0 pt+0 / −0
Serves /llms.txt
of 518
23%+0.8 pt+4 / −0
Serves /agents.md
of 518
3.1%0 pt+0 / −0
Publishes an agent-skills index
of 518
2.5%-0.2 pt+0 / −1
Publishes an MCP server card
of 518
2.5%0 pt+1 / −1
Publishes an API catalog (RFC 9727)
of 518
2.5%+0.2 pt+1 / −0
Publishes an A2A agent card
of 518
0.6%0 pt+0 / −0
Declares any agent capability (any of the four above)
of 518
3.3%0 pt+1 / −1
At least one AI identity is refused (non-2xx) where the browser got 2xx
of 518
18%-1 pt+2 / −7
At least one AI identity gets under 70% of the browser's bytes (both 2xx)
of 518
3.9%+0.2 pt+3 / −2
Names an MCP endpoint (server card, llms.txt, agents.md, a homepage link, or /mcp answers like one)
of 518
10.6%-0.2 pt+0 / −1
An MCP endpoint completes the handshake (initialize and tools/list; no tool called)
of 518
1.4%0 pt+0 / −0
Of sites naming an MCP endpoint, one completes the handshake or asks for credentials
of 55
78.2%0 pt+0 / −0
The A2A agent card names an endpoint url
of 518
0.4%0 pt+0 / −0
Of A2A cards naming a url, the url answers (any non-5xx, not an HTML page)
of 2
50%0 pt+0 / −0
Homepage carries WebMCP (navigator.modelContext, a WebMCP bridge, or an mcp+json block)
of 518
0.2%0 pt+0 / −0
Listed by Google Safe Browsing (all or some pages): browsers show a red warning
of 319
1.3%——
Listed as malware or phishing by a public malware-blocking DNS resolver
of 532
0%——
On the browsers' HSTS preload list (HTTPS-only from the first visit)
of 532
17.5%——
Homepage loads scripts or styles from another domain
of 532
62.2%——
Homepage loads scripts or styles from 5 or more other domains
of 532
6.4%——
Pins at least one outside script or style with integrity=
of 331
8.8%——
Loads a library from a public CDN without integrity=
of 532
9%——
Still loads scripts from a CDN taken over in 2024 (polyfill.io and its sister CDNs)
of 532
0%——
Homepage shows hidden spam links or cryptominer code
of 532
0%——
GPTBot refused where the browser got 2xx
of 518
15.3%-0.6 pt+2 / −5
ClaudeBot refused where the browser got 2xx
of 518
14.9%-0.4 pt+4 / −6
OAI-SearchBot refused where the browser got 2xx
of 518
10.2%-0.6 pt+2 / −5
PerplexityBot refused where the browser got 2xx
of 518
13.7%+0.4 pt+8 / −6

top 10,000, daily

Run 26 (finished 2026-11-22), 9776 sites, compared with run 19 (2026-11-15).

Biggest moves: PerplexityBot refused where the browser got 2xx +1.1 pt (88 gained, 26 lost); At least one AI identity is refused (non-2xx) where the browser got 2xx -0.3 pt (48 gained, 66 lost); robots.txt names the AI-training policy token (__usage_ai-training, IETF AIPREF draft) +0.3 pt (14 gained, 0 lost); Of sites naming an MCP endpoint, one completes the handshake or asks for credentials -0.3 pt (0 gained, 1 lost); Homepage answers CrawlCheck's census fetch (2xx after redirects) -0.2 pt (148 gained, 166 lost).

FigureThis weekChange, same sitesSites +/−
Homepage answers CrawlCheck's census fetch (2xx after redirects)
of 9776
57.8%-0.2 pt+148 / −166
Homepage refuses CrawlCheck's census fetch (400, 401, 403 or 429): blocked or rate-limited, not down
of 9776
13.9%+0.2 pt+95 / −78
No website answers at that name (5xx, 530, timeout or no answer)
of 9776
26.3%0 pt+83 / −79
Serves a robots.txt (200, not HTML)
of 5653
81.9%0 pt+29 / −27
robots.txt names a sitemap
of 4627
70.3%+0.1 pt+9 / −4
robots.txt carries a Content-Signal line
of 4627
6.5%+0.1 pt+6 / −1
Content-Signal says ai-train=no
of 4627
5%+0.1 pt+4 / −1
Content-Signal says ai-input=no or search=no
of 4627
0.5%0 pt+0 / −1
robots.txt names the AI-training policy token (__usage_ai-training, IETF AIPREF draft)
of 4623
1.1%+0.3 pt+14 / −0
robots.txt states a preference against AI training on / with that token
of 4623
0.6%+0.2 pt+10 / −0
robots.txt closes / to at least one of the 12 AI agents
of 4627
22.1%+0.1 pt+7 / −4
robots.txt closes / to GPTBot
of 4627
16.2%0 pt+4 / −5
robots.txt closes / to OAI-SearchBot
of 4627
6.3%0 pt+3 / −1
robots.txt closes / to ChatGPT-User
of 4627
8%0 pt+2 / −2
robots.txt closes / to ClaudeBot
of 4627
15%0 pt+4 / −4
robots.txt closes / to Claude-SearchBot
of 4627
6%+0.1 pt+4 / −1
robots.txt closes / to PerplexityBot
of 4627
9%0 pt+1 / −2
robots.txt closes / to Google-Extended
of 4627
13.4%-0.1 pt+1 / −4
robots.txt closes / to Applebot-Extended
of 4627
13.3%0 pt+4 / −4
robots.txt closes / to CCBot
of 4627
17.2%0 pt+4 / −3
robots.txt closes / to Bytespider
of 4627
17.1%0 pt+5 / −3
robots.txt closes / to Meta-ExternalAgent
of 4627
14%+0.1 pt+8 / −4
robots.txt closes / to Amazonbot
of 4627
13.7%0 pt+6 / −4
Serves /llms.txt
of 5653
17%+0.1 pt+13 / −8
Serves /agents.md
of 5653
1.5%0 pt+3 / −4
Publishes an agent-skills index
of 5653
1.1%0 pt+0 / −1
Publishes an MCP server card
of 5653
1.2%0 pt+1 / −1
Publishes an API catalog (RFC 9727)
of 5653
1.3%0 pt+1 / −1
Publishes an A2A agent card
of 5653
0.3%0 pt+0 / −2
Declares any agent capability (any of the four above)
of 5653
1.6%0 pt+1 / −1
At least one AI identity is refused (non-2xx) where the browser got 2xx
of 5653
22.4%-0.3 pt+48 / −66
At least one AI identity gets under 70% of the browser's bytes (both 2xx)
of 5653
2.1%-0.1 pt+8 / −12
Names an MCP endpoint (server card, llms.txt, agents.md, a homepage link, or /mcp answers like one)
of 5653
6%+0.1 pt+9 / −6
An MCP endpoint completes the handshake (initialize and tools/list; no tool called)
of 5653
1%0 pt+1 / −0
Of sites naming an MCP endpoint, one completes the handshake or asks for credentials
of 338
78.1%-0.3 pt+0 / −1
The A2A agent card names an endpoint url
of 5653
0.2%0 pt+0 / −0
Of A2A cards naming a url, the url answers (any non-5xx, not an HTML page)
of 9
44.4%0 pt+0 / −0
Homepage carries WebMCP (navigator.modelContext, a WebMCP bridge, or an mcp+json block)
of 5653
0.1%0 pt+0 / −0
Listed by Google Safe Browsing (all or some pages): browsers show a red warning
of 4143
0.8%——
Listed as malware or phishing by a public malware-blocking DNS resolver
of 5674
0.3%——
On the browsers' HSTS preload list (HTTPS-only from the first visit)
of 5671
8%——
Homepage loads scripts or styles from another domain
of 5675
66.9%——
Homepage loads scripts or styles from 5 or more other domains
of 5675
10.5%——
Pins at least one outside script or style with integrity=
of 3796
9.9%——
Loads a library from a public CDN without integrity=
of 5675
12.8%——
Still loads scripts from a CDN taken over in 2024 (polyfill.io and its sister CDNs)
of 5675
0.1%——
Homepage shows hidden spam links or cryptominer code
of 5675
0%——
GPTBot refused where the browser got 2xx
of 5653
19.1%0 pt+38 / −39
ClaudeBot refused where the browser got 2xx
of 5653
19.9%+0.2 pt+66 / −55
OAI-SearchBot refused where the browser got 2xx
of 5653
12.4%-0.1 pt+24 / −31
PerplexityBot refused where the browser got 2xx
of 5653
16.7%+1.1 pt+88 / −26

Each frame compares the last run that finished in this ISO week (UTC) with the last run that finished before it. Change is like-for-like: points = (sites gained - sites lost) / sites compared, counting only sites measured in both runs and inside the figure's denominator both times, so a site that did not answer in one run cannot move a figure. points_all is the plain difference of the two percentages, and denominator_entered / denominator_left count the sites that moved in or out of a figure's base (often a site that did not answer one run). No site is named; per-site records: https://crawlcheck.io/api/census/records.

Signed JSON: /api/v1/census/weekly?week=2026-W47 · feed: Atom · method: /api/census. Editions: 2026-W46 · 2026-W45 · 2026-W44 · 2026-W43 · 2026-W42 · 2026-W41 · 2026-W40.

Check it yourself

Every number here comes from the signed JSON at /api/v1/census/weekly; each copy carries its own sha256 and an Ed25519 signature by key 1hHmE3fo1rg5…. Verify it with no call to us:

node crawlcheck-verify.mjs --online census-weekly.json

The verifier: /crawlcheck-verify.mjs, zero dependencies.

Editions

One sealed edition per ISO week, never edited. Feed: Atom · all instruments: /api/v1/instruments.