How websites open up to AI agents

Every month we check hundreds of thousands of the web's largest sites for the files and endpoints AI agents use to find content, APIs and tools, from llms.txt to MCP servers. A share of them is checked each day, so the page is updated daily.

16%

of sites publish at least one agent standard

llms.txt

is the most adopted agent standard, on 12% of sites

1343

sites expose an MCP server for agents

agent and API standards on connect.com, forter.com, more than any other site

Adoption by standard

Share of sites where the standard is present and well-formed. Select a row to see how sites announce it and which sites support it, and a site to open the evidence.

Agent standards

Files and endpoints made for AI agents.

llms.txtA Markdown summary of the site with links, written for language models. 12%108,909 sites

+108899

How the 108,909 sites were found

/llms.txt108,722
/llms-full.txt29,966
<link> tag459
Link header443

30,222 sites were found more than one way, so the counts add up to more than 108,909.

Empty or placeholder file only: autode.ro, eumeworld.com, promarineantioxidants.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, butane-karaj.com, casapark.fr, cassidytravel.ie, catholicscientists.org, centreforentrepreneurs.org, changenow.world, discoverqatar.qa, donateursbelangen.nl, easyenarm.com, flightcatchers.com, garciaenarm.com, hip.agency, internationaldays.co, la-maison-bleue.fr, lumenchristi.org, nacesports.org, ncse.ie, negahhospital.com, nobetcigazete.com, olejeobchod.cz, pioneerfund.vc, ppmarket.fi, proprietesforestieres.com, qatarairwaysholidays.com, rajved.in, rhagallery.ie, scireproject.com, scnnewsindia.com, snis.ch, socialjusticebooks.org, travelbyinspire.co.uk, 123kolo.cz, 123led.sk, 2jakost.cz, 365shop.cz, 3market.cz, 3v1trivjednom.cz, 4kids.cz, 4window.cz, ababu.cz and 3,135 more

UCPA Universal Commerce Protocol profile that lets agents shop and check out. 6.1%58,017 sites

+58007

How the 58,017 sites were found

/.well-known/ucp57,888
Link header11,900
<link> tag11,522

11,840 sites were found more than one way, so the counts add up to more than 58,017.

Empty or placeholder file only: overskill.com, crowcrowcrow.com, zr.ru, deployhq.com, izomat.cz, alcohome.co.il, atschool.com.ua, bracha-online.co.il, databuddy.cc, deployhq.com, luxuryproperty.com, posteverywhere.ai, shopup.com, gtopala.com, saunat.fi, tacdane.dk, life360.com, moodyradio.org, otp24hr.com, paypal.ai, sanwapearl.com.hk, totm.com, acdirect.co.za, bitter.co.il, bollekesfeest.be, dirtybirdusa.com, dreamhomeil.com, edcwsxqaz.site, ff14restanet.com, fjallraven.jp, gayseniorsexdate.com, iqos.ro, jeffery-west.co.uk, kt-pay.com, nemlig.com, nobetcigazete.com, py.pl, seniorsexpartner.com, sportspirit.md, stylion.pl, veryfi.com, yrke.pl, ag.pl, appinformatica.com, arturo-obuwie.pl, aspena24.pl, atriumshop.eu, audio24.pl, beanbox.com, beep.es and 59 more

AGENTS.mdInstructions for AI agents, in the format used by coding agents. 2.1%19,589 sites

+19579

Empty or placeholder file only: cozyla.com, eumeworld.com, promarineantioxidants.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, shopping.mattel.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, paloaltonetworks.com, polizeiticker.ch, prismacloud.io, alan.com, alpinecredits.ca, jepang.org, paloaltonetworks.ca, paloaltonetworks.co.uk, paloaltonetworks.com, paloaltonetworks.com.au, paloaltonetworks.de, paloaltonetworks.es, paloaltonetworks.fr, paloaltonetworks.in, paloaltonetworks.jp, polovniautomobili.com, reading.com, topol.io, 123winzone.biz, 222rsn5.com, 333v-chat.com, 33bmw247onlinecustomerservice.com, 33hub-game.xyz, 52dcs.com, 53s.live, 55rkcs.com, 55rtcs.com, 5n5ncs.com, 69jlq.site and 189 more

Markdown pagesPages that link to a clean Markdown version of themselves. 0.2%2,303 sites

+2293

How the 2,303 sites were found

<link> tag1,909
Link header855

461 sites were found more than one way, so the counts add up to more than 2,303.

Agent SkillsAn index of skills (packaged instructions) that agents can load. 0.2%1,622 sites

+1612

Empty or placeholder file only: fischersports.com, turboweb.co.nz, digital.ai, digital.ai, digital.ai, digital.ai, digital.ai, digital.ai, eachlabs.ai, opencourtdata.uk, uncensored.chat, hammerai.com, hilamat.com, kysely.dev, selia-tech.com, tinfoil.sh, trunk.io, xrpl.org, yarooms.com, codehelp.in, imker.pl, imker.pl, rwsdk.com, rwsdk.com, techritual.com, kafaratplus.com, kick-video.download, projector-audio.co.il, zbengfashion.co.il, moodyradio.org, paypal.ai, acdirect.co.za, alps2alps.com, anntaylor.com, anuhar.com, apollographql.com, apollographql.com, edcwsxqaz.site, ff14restanet.com, fjallraven.jp, gayseniorsexdate.com, ipinfo.io, iqos.ro, kt-pay.com, nationstates.net, nobetcigazete.com, py.pl, seniorsexpartner.com, skyzone.com, mfn.se and 24 more

MCPA Model Context Protocol server that gives agents tools and data. 0.1%1,343 sites

+1333

How the 1,343 sites were found

/mcp1,207
/mcp/server-card88
Link header54
<link> tag40
/.well-known/mcp-server31
/mcp/server-card.json11

81 sites were found more than one way, so the counts add up to more than 1,343.

Empty or placeholder file only: thorchain.org, apis.io, justidea.agency, gilibookings.com, mediccentre.pl, pixeldojo.ai, seychellesbookings.com, 3ddevice.com.ua, joost.blog, joost.blog, lolahealth.com, panwybierak.pl, coreaesthetics.com.au, filebin.ca, imagebin.ca, orpc.dev, pastebin.ca, pinecone.io, smollaunch.com, turl.ca, apptoto.com, craftwork.design, crewai.com, edenspiekermann.com, fraus-klett.cz, ibin.co, mcrender.net, opentweet.io, survivejs.com, tracextech.com, trmnl.com, trmnl.com, wave.co, webscraping.ai, anyvan.com, cijenegoriva.ba, codex-reset.com, csboard.com, dogechain.com, fastcomments.com, fliplet.com, inside.ad, kasada.io, lejebolig.dk, ryor.cz, senesi.cz, suitedash.com, upcomers.com, wpsubscription.co, arcblock.io and 748 more

ai.txtSpawning's ai.txt policy on how AI may use the site's content. 0.1%1,058 sites

+1048

Empty or placeholder file only: amanha.com.br, desirescans.com, drakecomic.net, drakecomic.net, jornaldebarueri.com.br, kaynscans.com, noveldex.io, valirscans.org, zr.ru, wafootball.com.au, att.com, websitespeedy.com, cloudshare.com, cyclr.com, lih.co.jp, diariodevaldivia.cl, docs.garden.io, docs.ockam.io, electronforge.io, futurimedia.com, futurimedia.com, grupoindustronic.com, guardz.com, ired.team, learnrxjs.io, libsodium.gitbook.io, mangoapps.com, mixo.io, reporteminero.cl, saashero.net, simcorner.com, songsterr.com, way2quran.com, air-dr.com, assuranceendirect.com, avensure.com, avenuez.com, bindi-online.com, boxdepo.it, centraleyes.com, chathub.gg, chatthing.ai, cognyte.com, comarchesklep.pl, d-id.com, divascans.org, fatosdesconhecidos.com.br, get.assembla.com, iguazio.com, kaynscans.com and 740 more

AI CatalogA catalog of the AI resources a site offers. 0.1%1,013 sites

+1003

Empty or placeholder file only: 3ecpa.com.sg, fischersports.com, zpf.pl, arabnews.com, codigosdediamantes.com, deutschesee.de, minuto30.com, mundoapprd.com, prostokorysno.com.ua, publitec.de, solar-haus.de, 3ecpa.com.my, akkushop-austria.at, arabnews.jp, brightstart.com, brightstart.com, demmer-shop.de, dihk-bildung.shop, dm-folien.com, elbsand.com, fluffcore.com, fluffcore.com, fluffcore.com, fruitoftheloom.eu, graf-dichtungen.de, iswari.com, kysely.dev, par71.de, raddios.com, repgalaxy.com, russellathletic.com, schuherlebnis.de, spalding.com, towebp.io, veikkaajille.com, wsu-industrials.com, zoundhouse.de, abicart.com, abicart.com, apostilasdominio.com.br, arabicescorts.com, besserdampfen.de, bimcontent.com, busnese.com, chatarg.com, cheaper-quotes.com, dertyunmklooplop.shop, emilfrey.de, fermeria.sk, ffnews.com and 64 more

A2AAn Agent2Agent card describing an agent that other agents can call. <0.1%454 sites

+444

How the 454 sites were found

/.well-known/agent-card.json440
Link header63
<link> tag6

54 sites were found more than one way, so the counts add up to more than 454.

Empty or placeholder file only: zr.ru, izomat.cz, farmaciagranvia216.es, paypal.ai, surepetcare.com, bollekesfeest.be, edcwsxqaz.site, ff14restanet.com, fjallraven.jp, gayseniorsexdate.com, imagy.app, iqos.ro, kt-pay.com, nobetcigazete.com, py.pl, seniorsexpartner.com, sylvansport.com, mfn.se, nownodes.io, redstone.finance, arbetsgivarverket.se, jobbastatligt.se, mountainwolf.com, opereta.hr, stitchfiddle.com, wellsfargoadvisors.com

AI pluginThe OpenAI plugin manifest, the first agent-discovery file (now retired). <0.1%416 sites

+406

How the 416 sites were found

/.well-known/ai-plugin.json413
Link header18
<link> tag10

25 sites were found more than one way, so the counts add up to more than 416.

Empty or placeholder file only: zr.ru, dils.pt, izomat.cz, moodyradio.org, paypal.ai, acdirect.co.za, bollekesfeest.be, edcwsxqaz.site, ff14restanet.com, fjallraven.jp, gayseniorsexdate.com, iqos.ro, kt-pay.com, nobetcigazete.com, northfitnessthailand.com, py.pl, seniorsexpartner.com, allentowndiocese.org, archmil.org, archny.org, b.sc, capcut.com, capcut.com, detroitcatholic.com, diocesefwsb.org, eztexting.com, mfn.se, nohrd.com, arbetsgivarverket.se, biblino.de, dateland.co.il, jobbastatligt.se, mountainwolf.com, opereta.hr, rusdate.co.il, rusdate.de, rusdate.net, rusdate.us, ukrdate.net, wellsfargoadvisors.com

WebMCP since 28 Sep 2026Tools a page declares for an in-browser agent to call, instead of clicking through the page. <0.1%283 of 744,767 sites checked so far

+273

Tracked since 28 Sep 2026. So far 744,767 of the 943,453 reachable sites have been checked for it, as each comes up for its monthly recrawl; the share is over those.

How the 283 sites were found

script call151
<form toolname>137

5 sites were found more than one way, so the counts add up to more than 283.

ARDAgentic Resource Discovery: an index of agent-facing resources. <0.1%279 sites

+269

How the 279 sites were found

/.well-known/ard.json265
<link> tag80
Link header39

93 sites were found more than one way, so the counts add up to more than 279.

Empty or placeholder file only: 3ecpa.com.sg, zpf.pl, codigosdediamantes.com, mundoapprd.com, prostokorysno.com.ua, 3ecpa.com.my, repgalaxy.com, veikkaajille.com, apostilasdominio.com.br, bimcontent.com, chatarg.com, dpbossnet.me, dpbossnet.me, fermeria.sk, ffnews.com, hundredburgers.com, practicatest.cl, practicatest.co, practicatest.com, practicatest.cr, roadtrippin.fr, sattamatka.day, zeusx.com, moodyradio.org, paypal.ai, acdirect.co.za, anntaylor.com, bollekesfeest.be, edcwsxqaz.site, ff14restanet.com, fjallraven.jp, gayseniorsexdate.com, iqos.ro, kt-pay.com, nobetcigazete.com, py.pl, seniorsexpartner.com, stelorder.com, mfn.se, arbetsgivarverket.se, badische-zeitung.de, badische-zeitung.de, jobbastatligt.se, mountainwolf.com, opereta.hr, wellsfargoadvisors.com

Poppy since 14 Oct 2026A Personal Agent Protocol profile that lets a user's personal agent sign in and complete tasks through the company's APIs or agent. <0.1%16 of 408,186 sites checked so far

Tracked since 14 Oct 2026. So far 408,186 of the 943,453 reachable sites have been checked for it, as each comes up for its monthly recrawl; the share is over those.

Empty or placeholder file only: zr.ru, izomat.cz, data.nasdaq.com, paypal.ai, ff14restanet.com, fjallraven.jp, iqos.ro, nobetcigazete.com, py.pl, arbetsgivarverket.se

agents.jsonA description of the flows and APIs an agent can use on the site. <0.1%28 sites

+18

Empty or placeholder file only: donotage.org, phishdestroy.io, zr.ru, monday.com, monday.com, amzscout.net, careerfaqs.com.au, izomat.cz, loi.com.uy, loichile.cl, mibelleza.uy, reducto.ai, smollaunch.com, zapier.com, zapier.com, cinema-timisoara.ro, iienstitu.com, opentweet.io, render.com, render.com, uludagkoleji.com, zapier.com, noorderlink.nl, podpage.com, podpage.com, shopsavvy.com, suitedash.com, tabak-brucker.de, teh.net, watanserb.com, wow-online.ir, adaptyvbio.com, desku.io, fungies.io, inworld.ai, no8.io, seo-textagentur.at, apis.vin, blanche.gr, creativepeptalk.com, data.nasdaq.com, droppe.com, esimo.io, eurodata.lt, fast.io, moodyradio.org, paypal.ai, wpadminify.com, acdirect.co.za, appetencyrecruitment.com.au and 63 more

Trust attestationMerchant trust or certification evidence that agents can verify. <0.1%17 sites

How the 17 sites were found

/.well-known/trust-attestation16
/.well-known/trust-issuer-metadata16
<link> tag0
Link header0

15 sites were found more than one way, so the counts add up to more than 17.

Empty or placeholder file only: otp24hr.com, sanwapearl.com.hk, acdirect.co.za, edcwsxqaz.site, kt-pay.com, nationstates.net, bloemencorso-bollenstreek.nl, chanakyya.com, hbveilen.nl, jubholland.nl, lined.nl, opereta.hr, wellsfargoadvisors.com

API & auth discovery

Machine-readable API and identity descriptions agents build on.

OAuth/OIDC server since 28 Sep 2026Where agents sign in and get tokens: OAuth server metadata (RFC 8414) or OpenID Connect discovery. 6.8%50,863 of 744,767 sites checked so far

+50853

Tracked since 28 Sep 2026. So far 744,767 of the 943,453 reachable sites have been checked for it, as each comes up for its monthly recrawl; the share is over those.

How the 50,863 sites were found

/.well-known/oauth-authorization-server48,455
/.well-known/openid-configuration22,155

19,747 sites were found more than one way, so the counts add up to more than 50,863.

Empty or placeholder file only: zr.ru, izomat.cz, xxl.fi, xxl.no, xxl.se, appsfactory.de, edenspiekermann.com, fraus-klett.cz, interactivewish.com, lawsociety.ly, boulderseomarketing.com, iletisim.gov.tr, lejebolig.dk, octolens.com, ryor.cz, senesi.cz, skdesu.com, unity-consulting.com, yutori.com, exclusivasex.com.br, koop.cz, wikitolica.com, workin.space, data.nasdaq.com, fashionnova.com, paragraph.com, paypal.ai, sanwapearl.com.hk, upway.at, upway.de, upway.fr, apkgame.com.pk, bollekesfeest.be, centuaryindia.com, ff14restanet.com, fjallraven.jp, gayseniorsexdate.com, iqos.ro, jeffery-west.co.uk, mepcobillchek.com, nemlig.com, nobetcigazete.com, py.pl, ramzanpackage8070.pk, seniorsexpartner.com, sliplane.io, soilfoodweb.com, wittysparks.com, 567zk.net, allcamsex.com and 92 more

OAuth resourceOAuth Protected Resource Metadata (RFC 9728), used to authorise MCP clients. 5.8%54,331 sites

+54321

Empty or placeholder file only: zr.ru, izomat.cz, ehn.org, ehn.org, gopenske.com, moodyradio.org, paypal.ai, rebelmouse.com, runpost.in, sanwapearl.com.hk, thecipherbrief.com, upway.at, upway.de, upway.es, upway.fr, upway.it, upway.nl, acdirect.co.za, adagio.com, amazing.today, anuhar.com, blooloop.com, bollekesfeest.be, brit.co, comicsands.com, contentbox.co.kr, edcwsxqaz.site, ff14restanet.com, fjallraven.jp, gayseniorsexdate.com, health-bank.co.kr, huwise.com, huwise.com, huwise.com, iqos.ro, jeffery-west.co.uk, kt-pay.com, louderwithcrowder.com, nationstates.net, nemlig.com, nobetcigazete.com, py.pl, researchers.kr, runrun.it, seniorsexpartner.com, allbusiness.com, belvarosiorvosicentrum.hu, ca.billboard.com, conservativereview.com, dailyclimate.org and 74 more

JWKSPublic signing keys, used to verify tokens and signed statements. 0.3%2,450 sites

+2440

Empty or placeholder file only: crowcrowcrow.com, webtures.com, harunyahya.info, bulletproofautomotive.com, cyclewalay.com, maker.co, pangram.com, swmansion.com, autoclub.it, blab.co, blab.co, dailybot.com, docuwriter.ai, free3d.io, free3d.io, leanlabs.com, leanlabs.com, motorsclub.com, novita.ai, nrwz.de, zambiajobalerts.com, zenity.io, contextual.ai, crewai.com, farmertofridge.com.au, hauteflorist.co.uk, heroui.com, heroui.com, imforza.com, intouchcx.com, kairatoon.com, levelsolve.com, luxuryproperty.com, maspormenos.net, meiro.io, midu.dev, mind-core.com, onceinteractive.com, palette.site, ui-skills.com, a-hentai.tv, anara.com, aquamagis.de, auamed.org, biwapp.de, bluewiremedia.com.au, bstok.pl, chachocarp.es, cloudshare.com, ctmusicshop.com and 208 more

API CatalogA list of the site's APIs at /.well-known/api-catalog (RFC 9727). 0.3%2,445 sites

+2435

How the 2,445 sites were found

/.well-known/api-catalog2,211
Link header1,756
<link> tag132

1,579 sites were found more than one way, so the counts add up to more than 2,445.

Empty or placeholder file only: trucksbuses.com, kafaratplus.com, spheron.network, fontainebleau.com, happyfox.com, moodyradio.org, paypal.ai, sanwapearl.com.hk, bollekesfeest.be, edcwsxqaz.site, ff14restanet.com, fjallraven.jp, gayseniorsexdate.com, iqos.ro, iwanbanaran.com, jeffery-west.co.uk, kt-pay.com, nationstates.net, nemlig.com, nobetcigazete.com, py.pl, seniorsexpartner.com, bestcoloringpages.ai, camdirect.fr, loft.com, yourmed.clinic, bloemencorso-bollenstreek.nl, chanakyya.com, globalincidentmap.com, hbveilen.nl, iconfont.cn, jubholland.nl, kingscross.co.uk, kiriminaja.com, lined.nl, opereta.hr, premieronline.com, topigeon.com, wellsfargoadvisors.com

OpenAPIA machine-readable description of the site's HTTP API. 0.2%1,840 sites

+1830

How the 1,840 sites were found

/openapi.json1,329
Link header735
<link> tag130

327 sites were found more than one way, so the counts add up to more than 1,840.

Empty or placeholder file only: orq.ai, zr.ru, rxresu.me, izomat.cz, cobot.me, ifttt.com, stedi.com, transloadit.com, checklyhq.com, ifttt.com, moodyradio.org, openrouter.ai, subscribr.ai, acdirect.co.za, aiven.io, campusaxis.pk, edcwsxqaz.site, ff14restanet.com, iqos.ro, kt-pay.com, nationstates.net, nazwa.pl, nobetcigazete.com, turningstone.com, wellnessliving.com, 68team.com, dartfreakz.nl, dartsnews.com, dutchcowboys.nl, f1maximaal.nl, fitorforestal.com, foot01.com, grandprix247.com, idlprocycling.com, laroma24.it, loft.com, mannenzaken.nl, netart.com, netart.com, netart.com, netart.com, pianetabasket.com, ruck.co.uk, showblad.nl, tennisuptodate.com, tv.team, u.today, voetbal24.be, westhamworld.co.uk, aenova-group.com and 67 more

Web basics

Long-established standards, for comparison.

robots.txtCrawler access rules (RFC 9309). 67%634,552 sites

+634542

Empty or placeholder file only: builtwith.com, cinema-timisoara.ro, alkarimfabric.com, anybody.hu, beetlesgel.com, bokksu.com, eastfork.com, eu.balibodyco.com, eyespysupply.com, huntparts.ca, juliettoys.com, mac-center.com.pe, maingear.com, myvega.com, pewabic.org, spicycubes.com, steelcollectibles.com, threadlogic.com, trymeloair.com, turtlewings.co.in, urbanvibesx.com, basisbeveiliging.nl, clapforourcarers.co.uk, fgo.ro, horosho-tam.ru, istyle.iq, momentousinstitute.org, slides.com, slides.com, aab.uk, adsanityplugin.com, aisingapore.org, aisler.net, app.kit.com, app.kit.com, applemagazine.com, artwinlive.com, audial.ir, avohq.io, avraread.com, babouti.ma, base-uk.org, bayerproperty.com, blogin.co, bv-osteopathie.de, callstack.com, caribbeanbiodiversityfund.org, catholiccharities.org, cccwale.com, chanhkien.org and 49,600 more

SitemapA list of the site's pages for crawlers. 50%469,830 sites

+469820

Empty or placeholder file only: zr.ru, nuezpecana.org, wpfusion.com, verein-naturschutzpark.de, webscraping.ai, biwapp.de, oceandrop.com.br, geoponiko-kentro.gr, goldex.sk, htmlweb.ru, iraklisoutdoor.gr, metorik.com, bibik.cz, chatthing.ai, comercialisabella.cl, futterattacke.de, gdandtbasics.com, goldex.cz, icook.tw, lovegraph.me, nailsmade.store, odpiralnicasi.com, offers.jp, openhours.es, qualialife.com, quintadb.com, saasbrowser.com, viqtory.com, webshopmanager.com, 1up-usa.com, 50-ml.de, 50-ml.it, africanfarming.com, agendasorocaba.com.br, agentic.market, anuhar.com, askbonus.com, benito.com, certifiedmtp.com, crashdebug.fr, dag-shg.de, dataart.com, doabooks.org, doppelgaenger.io, durable.com, durable.com, ecigmafia.com, ecolecamondo.fr, ecuhealth.org, ecuhealth.org and 2,956 more

Schema.orgStructured data (JSON-LD) on the homepage. 35%332,312 sites

+332302

Who supports what

The top 1,000 of the 177,886 sites that publish at least one agent or API standard, ranked by how many.

adopted empty or placeholder not checked: blocked by the site not found

Site	Standards	llms.txt	UCP	AGENTS.md	Markdown pages	Agent Skills	MCP	ai.txt	AI Catalog	A2A	AI plugin	WebMCP	ARD	Poppy	agents.json	Trust attestation	OAuth/OIDC server	OAuth resource	JWKS	API Catalog	OpenAPI	robots.txt	Sitemap	Schema.org
connect.com	14
forter.com	14
pedagioeletronico.com.br	13
youspot.com	13
donotage.org	12
sitegpt.ai	12
whisperchat.ai	12
yruz.one	12
bibigpt.co	11
ora.ai	11
overskill.com	11
telnyx.com	11
thorchain.org	11
apis.io	10
brandfetch.com	10
context.dev	10
context.dev	10
crowcrowcrow.com	10
deathbycaptcha.com	10
echowave.io	10
emorahealth.com	10
finsi.ai	10
froggyads.com	10
justidea.agency	10
littlebirdelectronics.com.au	10
muumimaailma.fi	10
podscan.fm	10
publica.la	10
razorpay.com	10
segura.security	10
spice.ai	10
suno.bi	10
ticketfairy.com	10
usama.dev	10
webtures.com	10
zenrows.com	10
amanha.com.br	9
answeringservicecare.com	9
apify.com	9
asdecarreaux.com	9
attrition.org	9
audiopod.ai	9
boutiquerugs.com	9
brandfetch.com	9
carregados.com.br	9
clashofcoins.com	9
desirescans.com	9
drakecomic.net	9
drakecomic.net	9
edicola.shop	9
eventify.io	9
flightqueue.com	9
flightseatmap.com	9
frase.io	9
getdex.com	9
gilibookings.com	9
h3manth.com	9
harunyahya.info	9
iacc.cl	9
intim.to	9
ippodotea.com	9
jornaldebarueri.com.br	9
kaynscans.com	9
layer.ai	9
leadmagic.io	9
mediccentre.pl	9
moon.chat	9
mstore.ie	9
noveldex.io	9
orq.ai	9
phishdestroy.io	9
pixeldojo.ai	9
quicknode.com	9
razorpay.com	9
razorpay.com	9
seychellesbookings.com	9
stingressos.com.br	9
surrealdb.com	9
tessa.tech	9
tuotroestanco.com	9
valirscans.org	9
zernio.com	9
zernio.com	9
zig.tickets	9
zr.ru	9
3ddevice.com.ua	8
4over4.com	8
accentuate.io	8
afun888.com	8
alser.kz	8
anahana.com	8
apideck.com	8
arcade.dev	8
autorfp.ai	8
bird.com	8
bird.com	8
bird.com	8
bird.com	8
boldsign.com	8
brale.xyz	8
breakcold.com	8
bulletproofautomotive.com	8
codetrendy.com	8
context7.com	8
cyclewalay.com	8
developers.cloudflare.com	8
developers.cloudflare.com	8
developers.cloudflare.com	8
divan.com.ua	8
docs.x.com	8
doctorsim.com	8
earthfirealliance.org	8
exa.ai	8
fly.io	8
formswrite.com	8
frsport.com	8
gecheckt.at	8
getmed.pl	8
getname.pl	8
getterms.io	8
gymbeam.ba	8
gymbeam.cz	8
gymbeam.hu	8
gymbeam.it	8
gymbeam.ro	8
gymbeam.rs	8
gymbeam.si	8
gymbeam.ua	8
helius.dev	8
homesage.ai	8
hostinger.com	8
hostinger.com	8
hulkapps.com	8
ibl.ai	8
ingracio.adv.br	8
joost.blog	8
joost.blog	8
kolayimar.com	8
lightinfitness.com	8
listenlabs.ai	8
lolahealth.com	8
maintenancec.com	8
maker.co	8
master-addons.com	8
monday.com	8
monday.com	8
mount-it.com	8
nas.com	8
necesit.ro	8
ngrok.com	8
ngrok.com	8
ngrok.com	8
nitrotiger.com	8
nouw.com	8
nuezpecana.org	8
open.money	8
ovidius-ch.ro	8
pachca.com	8
padlet.com	8
padlet.com	8
pangram.com	8
panwybierak.pl	8
poolcomet.com	8
pressefeuer.at	8
rxresu.me	8
savvycal.com	8
scrolllaunch.com	8
semicvetic.com	8
shakudo.io	8
skillsmp.com	8
specification.website	8
swmansion.com	8
trebam.hr	8
usercentrics.com	8
visbook.com	8
winpot.mx	8
worldmonitor.app	8
zuplo.com	8
3byggetilbud.dk	7
aaronknight.com.au	7
abaton.es	7
airbyte.com	7
alphasignal.ai	7
amzscout.net	7
aptos.dev	7
argos-ci.com	7
arms-bg.com	7
asuperiormechanical.com	7
autoclub.it	7
azurally.com	7
batterytrader.com	7
bazel.build	7
bazel.build	7
bird.com	7
blab.co	7
blab.co	7
blogseo.io	7
boei.help	7
boldbi.com	7
bresser.com	7
bresser.fr	7
browserless.io	7
buscamed.com	7
capafy.ai	7
careerfaqs.com.au	7
centraldeconcursos.com.br	7
ciclopoli.de	7
clerk.com	7
clickclickmedia.com.au	7
clickhouse.com	7
clueso.io	7
codeage.com	7
codesiya9a.ma	7
confident-ai.com	7
contactmonkey.com	7
coreaesthetics.com.au	7
crctransport.us	7
cyber-securite.fr	7
dailybot.com	7
dash0.com	7
degraucultural.com.br	7
deployhq.com	7
developer.nylas.com	7
dils.pt	7
document360.com	7
document360.com	7
docuwriter.ai	7
dulotec.bg	7
dupe.com	7
dynamicbusiness.com	7
dynamicbusiness.com	7
elmin7a.com	7
every.to	7
farmacbd.es	7
filebin.ca	7
fish.bg	7
fly.io	7
free3d.io	7
free3d.io	7
frikko.com	7
getlago.com	7
gonitrafik.ru	7
gymbeam.at	7
gymbeam.com	7
gymbeam.de	7
gymbeam.gr	7
gymbeam.hr	7
gymbeam.pl	7
gymbeam.sk	7
had.si	7
haveibeenpwned.com	7
hiskingdomprophecy.com	7
host-tracker.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
hostinger.com	7
ibest.tw	7
imagebin.ca	7
insst.es	7
interlink.ro	7
izomat.cz	7
jobicy.com	7
joshwoodcolour.com	7
kevurugames.com	7
kraess.dk	7
lad.lviv.ua	7
leanlabs.com	7
leanlabs.com	7
legitgamblingsites.io	7
lelo.com	7
listnr.ai	7
listnr.ai	7
llmpulse.ai	7
lobehub.com	7
lobsteranywhere.com	7
loi.com.uy	7
loichile.cl	7
lokmanhekim.com.tr	7
loops.so	7
lottenypalace.com	7
lovie.co	7
maldivesindependent.com	7
manycontacts.com	7
marketbetter.ai	7
mibelleza.uy	7
michelfortin.com	7
mindstamp.com	7
mishkan-hatchelet.co.il	7
momentic.ai	7
motaword.com	7
motorsclub.com	7
muapi.ai	7
mycvcreator.com	7
netlify.com	7
ngrok.com	7
ngrok.com	7
ngrok.com	7
nitnemsahib.com	7
novita.ai	7
nrwz.de	7
nuditok.com	7
orpc.dev	7
overclocking.com	7
pastebin.ca	7
peptidecritic.com	7
pinecone.io	7
prescientai.com	7
productlift.dev	7
pulsetic.com	7
pydantic.dev	7
railway.com	7
railway.com	7
reducto.ai	7
refunder.pl	7
refunder.se	7
renoise.ai	7
reservadirecto.com	7
rhombus.com	7
rossmanngroup.com	7
sanatory.ru	7
scalar.com	7
scalar.com	7
scrapfly.io	7
scrollerspub.com	7
searchbloom.com	7
sequenzy.com	7
skins.nl	7
smollaunch.com	7
speakai.co	7
sshh.io	7
supabase.com	7
supabase.com	7
supabase.com	7
supabase.com	7
supermemory.ai	7
superset.sh	7
thearmchairtrader.com	7
thesportsdb.com	7
timdeschryver.dev	7
tlapnet.cz	7
torial.com	7
tuist.dev	7
tuist.dev	7
turl.ca	7
unorouter.com	7
untitledui.com	7
vbout.com	7
vercel.com	7
vercel.com	7
vercel.com	7
vercel.com	7
vercel.com	7
vercel.com	7
vercel.com	7
vercel.com	7
wafootball.com.au	7
warmlyyours.com	7
warnermusic.fr	7
wpformation.com	7
wpfusion.com	7
xata.io	7
xxl.fi	7
xxl.no	7
xxl.se	7
yemenilaw.com	7
zambiajobalerts.com	7
zapier.com	7
zapier.com	7
zenity.io	7
zipwp.com	7
zoukgrouplv.com	7
21st.dev	6
3saf.com	6
7lle.com	6
aani-dani.com.sa	6
abadystore.com	6
abahanavillas.com	6
abchome.com	6
abounayefhoney.com	6
acmsnack.com	6
actan.com	6
activecampaign.com	6
activecampaign.com	6
adel-online.com	6
adwaastore.sa	6
adyarjewellery.com	6
affonso.io	6
agentclientprotocol.com	6
agentskills.io	6
agentskills.io	6
airbyte.com	6
aiterms.tw	6
akkushop.de	6
al-hazmi.net	6
al-jazirah.org	6
alammarigold.com	6
alamoudi-alize.com	6
alarrab.com	6
alasadiah.com	6
alborgdx.com	6
alcohome.co.il	6
alesaei-aes.com	6
alfarahgames.com	6
algharbis.com	6
alhabibshop.com	6
alhadayacenter.com	6
alhawwaj.com	6
alhazzaaautoparts.com	6
alhussainattara.com	6
allcompetitionclasses.com	6
almahmadispareparts.com	6
almas4perfumes.com	6
almusbahperfume.com	6
almuttasil.co	6
aloyayri.sa	6
alqalaavape.com	6
alsannat.com	6
alshamel.cc	6
alshiaka.com	6
alsultan-ksa.com	6
alsweed.sa	6
altuwijri.com	6
alwan7.com	6
amatizambiri.com	6
andrewconnell.com	6
anticheat.ac	6
apkcube.com	6
apos.audio	6
appsfactory.de	6
apptoto.com	6
arkanco.sa	6
armadaone1.com	6
artie.com	6
asdautoparts.com	6
askcory.ai	6
asrar-co.com	6
atbbq.com	6
atschool.com.ua	6
att.com	6
attio.com	6
axl.tech	6
b1care.com	6
backcare.com.sa	6
baiebrassage.fr	6
banafaforoud.com	6
baro-toys.com	6
batonia.com	6
baytalahidhia.com	6
becarestore.com	6
best-shield.com	6
bestprice.gr	6
bestself.co	6
beytna.store	6
bgpkit.com	6
blancoparfumes.com	6
blendcommerce.com	6
blink.new	6
bogenia.com.sa	6
bolaesportiva.com	6
boldreports.com	6
boleto7.com	6
bonsai.de	6
bracha-online.co.il	6
bravocenters.com	6
bresser.de	6
bresser.es	6
bresser.it	6
bresserbenelux.nl	6
buddy.works	6
builtwith.com	6
caltpro.com	6
candycartonsa.com	6
candyniche.com	6
carvek-ksa.com	6
casing7.com	6
cezesco.cz	6
chatboxai.app	6
choco-senjab.com	6
chytryrejstrik.cz	6
cinema-timisoara.ro	6
clarahair.com	6
cleandishes1.com	6
clickhouse.com	6
close.marketing	6
cloudflare.com	6
cloudflare.com	6
cloudflare.com	6
cloudflare.com	6
cloudflare.com	6
cloudflare.com	6
cloudflare.com	6
cloudflare.com	6
cloudflare.com	6
coffeesouq1.com	6
coinapi.io	6
coinpaprika.com	6
comparethecloud.net	6
contextual.ai	6
conversion.com.br	6
cookieyes.com	6
coolset.com	6
coravin.com	6
cozmazone.com	6
craftwork.design	6
crewai.com	6
cut.pro	6
cutoutandkeep.net	6
cyndra.ai	6
daralamirat.com.sa	6
darlena.com	6
darsonbol.com	6
databuddy.cc	6
de.gilson.com	6
deer-store.com	6
deployhq.com	6
desarksa.com	6
deshdashtii.com	6
developer.nvidia.com	6
dimers.com	6
dj.studio	6
docs.ag-ui.com	6
docs.digitalasset.com	6
dokan-iphone.com	6
dometrain.com	6
dreamlit.ai	6
drishtisignals.in	6
driveway.sa	6
dukan.me	6
durtalmsaj.com	6
edenspiekermann.com	6
eggdesign.ie	6
empress-clothing.com	6
en.paulmann.com	6
encontreumnerd.com.br	6
eseven-store.com	6
esiea.fr	6
evan2.com	6
exbna.com	6
exceeronline.com	6
exoplanetsa.com	6
explorium.ai	6
factory-moon.com	6
farmertofridge.com.au	6
fastbost.com	6
fayafidirati.com	6
fdeshdashti.com	6
firmy.ai	6
fixero.com	6
fomashop.com	6
fonteynspas.com	6
fontvibe.ai	6
fraus-klett.cz	6
fundraiseup.com	6
fyfaa.com	6
gamesekaio.com	6
gamitisa.com	6
gauravtiwari.org	6
genshin-builds.com	6
georgina-fashion.com	6
getcargo.ai	6
getcargo.ai	6
getforexbro.com	6
getlasso.co	6
getlasso.co	6
ghaydaamedical.com	6
gitbook.com	6
gkrawaiegeneva.com	6
glif.app	6
godo.is	6
gofastmcp.com	6
gogs.io	6
gonimedia.ru	6
gpt.space	6
gptr.dev	6
grabient.com	6
hauteflorist.co.uk	6
hebrsa.com	6
hedeya.com	6
henna-raghad.com	6
here.now	6
heroui.com	6
heroui.com	6
hexclave.com	6
hikerapi.com	6
hind-2026.com	6
hittaomstallningsorg.se	6
hovercode.com	6
htmlcsstoimage.com	6
htmlcsstoimage.com	6
htmlpub.com	6
huggingface.co	6
huggingface.co	6
huggingface.co	6
huggingface.co	6
huggingface.co	6
hyperleap.ai	6
ibin.co	6
iblackphone.com	6
igorslab.de	6
iienstitu.com	6
imforza.com	6
insst.es	6
interactivewish.com	6
interserver.net	6
intouchcx.com	6
iot-analytics.com	6
jens.marketing	6
jomlah.app	6
jsonschemaapp.com	6
judepack.com	6
juicer.io	6
juicyvapesa.com	6
k2n1.com	6
kadimhukuk.com.tr	6
kaffary.com	6
kairatoon.com	6
kajawal.com	6
karwa2030.com	6
kashkhastyle.com	6
kayanabaya.com	6
khalidhalabi.com	6
khatoonjewelry.com	6
knzalshita.com	6
kommunity.com	6
koorabreak.com	6
kznstore.com	6
laftah.com	6
lahzatzuhur.com	6
lamsethadaf.com	6
laratranslate.com	6
latitude.so	6
lawsociety.ly	6
leaderschairs.com	6
leadpages.com	6
leadpages.com	6
lebonheurperfumes.com	6
levelsolve.com	6
lgrworld.com	6
lifx.com	6
litlookzstudio.com	6
livecoins.com.br	6
livefresh.de	6
livekit.com	6
livnightclub.com	6
llamagen.ai	6
lmco1.com	6
luxuryproperty.com	6
madeinpowr.com	6
madinagolds.com	6
madinajewelleries.com	6
magista.store	6
malqurashisa.com	6
mamamia.com.au	6
mansorah.com	6
maraken-alshaima.com	6
martie.com	6
masar-bicycle.com	6
masarat-four.com	6
maspormenos.net	6
maxwarehouse.com	6
mcrender.net	6
medicube.sa	6
medrar.sa	6
meiro.io	6
melen.sa	6
mentimeter.com	6
mftackle.com	6
midu.dev	6
mieuxquedesfleurs.com	6
mind-core.com	6
miniso.sa	6
minty.com	6
minty.com	6
misspinkyy.com	6
mixersa.com	6
mleha.com	6
modelcontextprotocol.io	6
modelcontextprotocol.org	6
monitorindonesia.com	6
motonline.com.br	6
mr-bounty.com	6
mtjrthaki.com	6
mybookeg.com	6
mygezor.com	6
mylifefitnessstore.com	6
nammlah.com	6
nas.com	6
nashigold.com	6
navattic.com	6
neeo-store.com	6
nekomics.by	6
neon.com	6
newp.fr	6
newstep.sa	6
nglr7.com	6
nicheksa.store	6
njdisa.com	6
nouvona.com	6
numbersprotocol.io	6
ocean67.com	6
okshopsa.com	6
ola.cv	6
onahole.com	6
onceinteractive.com	6
ooopenlab.cc	6
opentweet.io	6
original-flask.com	6
otoparcasan.com	6
outletpharmacyonline.sa	6
oyuneks.com	6
palette.site	6
parclick.com	6
parsacoffee.com	6
patchworks.io	6
perfumechance.com	6
pharmabrand.sa	6
places.je	6
planefox.com	6
point-pharmacies.com	6
pointfive.co	6
pointoneonline.com	6
posteverywhere.ai	6
primocreno.com	6
proofcar1.com	6
propaideals.co.uk	6
prospyrmed.com	6
punkride.com	6
pvp-pc.com	6
qaydao.com	6
qssun.store	6
raffol.sa	6
rains.com	6
ralqurashi.com	6
rasees.net	6
realoficial.com.br	6
recurpost.com	6
render.com	6
render.com	6
requesty.ai	6
reservbookingengine.com	6
romansr.com	6
roouckan.com	6
rosa-clara.com	6
rosental.de	6
rsifir.com	6
rwnza.com	6
ryaheen.com	6
s-trx.com	6
sa.gentooshop.com	6
sa.osmaperfumes.com	6
sadir-sa.com	6
saeel.sa	6
salhea.com	6
sapling.ai	6
sara-makeup.com	6
saudipartsstore.com	6
sbaik.sa	6
sdmksa1.com	6
season.sa	6
seatpin.com	6
sedrahwear.com	6
sellerassistant.app	6
sendcloud.dev	6
seojuice.com	6
seoservices.com.br	6
serdababaya.com	6
seven-side.com	6
sfaqat.com	6
shaden1.com	6
shapermint.com	6
sharkgeming.com	6
sheikhmt.com	6
sherlockproject.xyz	6
shmbeautysa.com	6
shopcircle.co	6
shopup.com	6
simr.ir	6
sirfrancis.com	6
site.pro	6
skipcalls.com	6
sleephigh.com	6
sleepxfactory.com	6
smart-center-system.com	6
smart.com.sa	6
smbstore.com.br	6
smry.ai	6
snaf.co	6
socialbu.com	6
socialchamp.com	6
socialchamp.com	6
sorbe.bg	6
spartacusxx.com	6
speakeasy.com	6
sportswear.sa	6
star-blox.com	6
starhomee.com	6
startuphub.ai	6
store-49.com	6
streamkap.com	6
sufyanalayyaf.com	6
sultan-magfory.com	6
sultanaleasl.com	6
sultansstone.com	6
sunnaalnahda.com	6
supabase.com	6
supply.co	6
surrati.sa	6
survivejs.com	6
swimoutlet.com	6
synapseindia.com	6
tafassell.com	6
tambourine.com	6
tamrirah.com	6
tavily.com	6
tawreedly.com	6
techaro.lol	6
technisat.de	6
technisat.de	6
tenderkart.in	6
testahelha1.com	6
tfaseel.sa	6
thearabianpost.com	6
thetaedgecloud.com	6
thewpminute.com	6
thnyan.com	6
threfa.com	6
tidyhq.com	6
tierarzt-karlsruhe-durlach.de	6
tik-tok.porn	6
tme9games.com	6
toutelathailande.fr	6
tracextech.com	6
treated.com	6
trend-box.de	6
trianglestoresa.com	6
trmnl.com	6
trmnl.com	6
troesser.de	6
truto.one	6
tryshdad.com	6
tumerfashion.com	6
turbo360.com	6
turbo360.com	6
tw93.fun	6
twitterviewer.net	6
twix-store.com	6
ui-skills.com	6
uludagkoleji.com	6
uniticket.com.br	6
unokeeper.com	6
unspam.email	6
vanilla.sa	6
vapemodesa.com	6
verein-naturschutzpark.de	6
victrion1.com	6
vijvercentrum.nl	6
virti.com	6
virtualticket.com.br	6
vlamoursa.com	6
voltebyk.in	6
voo.nl	6
voxies.io	6
wafi4.com	6
wareefsa.com	6
wave.co	6
weaverse.io	6
webscraping.ai	6
websitespeedy.com	6
wesamsa.com	6
wherobots.com	6
whitefriday.sa	6
wholebeansa.com	6
wildmart.md	6
williampitt.com	6
windowsforum.com	6
withallo.com	6
wjhtektelecome.com	6
workiz.com	6
wtr.sa	6
y.uno	6
yellowqts.com	6
yelomark.com	6
you.com	6
youth4work.com	6
zapier.com	6
zarafaksa.com	6
zenmux.ai	6
zeymedya.com	6
zincoint.com	6
zoukgroup.com	6
zwembad.shop	6
007store.com	5
032c.com	5
0800donrouch.com.ar	5
1001coques.fr	5
100ampere.com	5
100percent.eu	5
100procenthardcore.com	5
100ways.com	5
101internet.id	5
10corsocomo.com	5
10keiya.com	5
10mois.com	5
11-11.in	5
111skin.com	5
11cart.com	5
11six24.com	5
1219avenue.com	5
12thtribe.com	5
1608wear.com	5
17th-st.com	5
1800tequila.com	5
1861.ca	5
18reasons.org	5
19crimes.com	5
1hairstop.in	5
1stphorm.com	5
200degs.com	5
20buckspin.com	5
20x200.com	5
21noticias.com	5
21overlays.com.au	5
21st-amendment.com	5
220-electronics.com	5
24-gute-taten.de	5
24-uurdeals.nl	5
247lingerie.co	5
24bottles.com	5
24hourcrafts.com	5
260samplesale.com	5
2baconil.com	5
2bglass.com	5
2cap.com.mx	5
2fastebikes.com	5
2foods.jp	5
2modern.com	5
2segypt.com	5
2tactic.com	5
2tall.com	5
2uul.com	5
2xu.com	5
300breaks.com	5
31philliplim.com	5
32degrees.com	5
33-mm.com	5
360cookware.com	5
365mashbir.co.il	5
3deksperten.dk	5
3dmatsusa.com	5
3max-sa.com	5
3sixteen.com	5
3t.bike	5
3wishes.com	5
42lolita.com	5
47brand.ca	5
47brand.co.jp	5
4akc.com	5
4chef.co.il	5
4dsystems.com.au	5
4elementos.com	5
4endurance.com	5
4endurance.de	5
4endurance.es	5
4endurance.pt	5
4feetshoes.com	5
4ocean.com	5
4phones.eu	5
4playessentials.co.za	5
4seating.com	5
5-bit.jp	5
500level.com	5
50plusmakeup.com	5
5150skateshop.com	5
5feet11.com	5
5hourenergy.com	5
5ktoys.com	5
5poundstuff.com	5
5starhookah.com	5
60sec.site	5
686.com	5
6dollarshirts.com	5
70cigars.com	5
70mai.store	5
72hours.ca	5
79bike.com	5

176,886 more sites publish fewer agent or API standards. 765,567 reachable sites publish no agent or API standard, and 78,484 were unreachable.

The standards

What each standard is for, where we look for it, and what an answer must look like to count.

Every file we fetch must pass these checks first

It is not the site's homepage: a file that redirects to the site's front page (/ with nothing after it) counts as not found. A redirect to the front page with a query string, such as /?page=sitemap, is a real answer and is checked like any other.
The answer, after following redirects, has a 2xx status. 401, 403 and 429 (login walls, bot protection, rate limits) are shown as not checked, as are paths the site's robots.txt disallows for us.
It is not the page the site returns for a random, nonexistent URL. The requested path is left out of that comparison, since many error pages repeat it, and so are lines that change on every request, such as a fetched_at: timestamp: a line whose value differs between two random URLs is ignored in all of them. If the site answers a random URL with an empty page, an empty file counts as not found rather than as an empty file.
It is not an HTML page where a file is expected.
The text files (ai.txt, llms.txt, llms-full.txt and AGENTS.md) are not an HTML page: they do not start like one, with a <!DOCTYPE html> in the first 256 bytes or a first line (after a template placeholder such as {#INCLUDE:head#}) that opens with <html>, <head>, <body>, <div> or <section>, also when encoded in a JSON string. An ai.txt holds no HTML markup at all in its first 4 KB, such as a closing </div> or </p> tag; llms.txt and AGENTS.md may, since they are often generated from a site's content. Tags inside Markdown code (inline code or code blocks) do not count, nor do opening tags alone, such as <br>, <img> or <link> mentioned in prose. They also do not start with Markdown front matter (a --- block) that reports an error status, such as status_code: 404: some sites answer a missing file with their error page rendered as Markdown.
It is not a JSON error message, whatever its status: a small JSON answer with an error code (401, 403, 404, 410), or a field saying the resource was not found, such as {"code": -1, "message": "ResourceNotFoundException"}.
Where JSON is expected, it is JSON: an empty or non-JSON body counts as not found, a file that starts as JSON but does not parse as invalid.
It has the structure the standard defines (below). Empty files, empty lists and placeholders such as {"notice": …} are shown separately and do not count as adopted.

Homepage signals come from <link> tags and Link headers only; a <meta> tag with the same name does not count.

llms.txt

A Markdown summary of the site with links, written for language models.

Counts when any of these is found:

/llms.txt: a non-empty text file that is not JSON, XML, an HTML or Markdown error page or a short error message such as 404 Not Found; the # Title heading the spec asks for is not required
/llms-full.txt: a non-empty text file that is not JSON, XML, an HTML or Markdown error page or a short error message such as 404 Not Found
a <link rel="llms"> or <link rel="llms-txt"> tag, or the same relation in a Link header, on the homepage

Specification →

UCP

A Universal Commerce Protocol profile that lets agents shop and check out.

Counts when any of these is found:

/.well-known/ucp: valid JSON: an object with a ucp field
a <link rel="ucp"> tag, or the same relation in a Link header, on the homepage

Specification →

AGENTS.md

Instructions for AI agents, in the format used by coding agents.

Counts when this is found:

/AGENTS.md: a non-empty text file that is not JSON, XML, an HTML or Markdown error page or a short error message such as 404 Not Found

Specification →

Markdown pages

Pages that link to a clean Markdown version of themselves.

Counts when this is found:

a <link rel="alternate" type="text/markdown"> tag, or the same relation in a Link header, on the homepage

Agent Skills

An index of skills (packaged instructions) that agents can load.

Counts when this is found:

/.well-known/agent-skills/index.json: valid JSON: an object with a list under skills (an empty list counts as empty)

Specification →

MCP

A Model Context Protocol server that gives agents tools and data.

Counts when any of these is found:

/.well-known/mcp-server: valid JSON: any object that is not a placeholder notice (the format is not settled yet)
/mcp: an MCP server's answer to a plain GET: an event stream, a JSON-RPC 2.0 message (but not with status 404), or a 401 whose WWW-Authenticate header points to OAuth resource_metadata
/mcp/server-card: valid JSON: an object with one of name or serverInfo
/mcp/server-card.json: valid JSON: an object with one of name or serverInfo
a <link rel="mcp"> or <link rel="mcp-server"> tag, or the same relation in a Link header, on the homepage

Specification →

ai.txt

Spawning's ai.txt policy on how AI may use the site's content.

Counts when this is found:

/ai.txt: a non-empty text file that is not JSON, XML, an HTML or Markdown error page or a short error message such as 404 Not Found, with at least one directive line Key: value, such as User-Agent: *, ai-train: disallow or Mokymas: ne; the keys of a leading Markdown front matter block (---) do not count, and binary data (a NUL byte) has none

Specification →

AI Catalog

A catalog of the AI resources a site offers.

Counts when this is found:

/.well-known/ai-catalog.json: valid JSON: an object with a list under entries (an empty list counts as empty)

A2A

An Agent2Agent card describing an agent that other agents can call.

Counts when any of these is found:

/.well-known/agent-card.json: valid JSON: an object with a name field and one of skills, url or supportedInterfaces
a <link rel="agent"> or <link rel="agent-card"> tag, or the same relation in a Link header, on the homepage

Specification →

AI plugin

The OpenAI plugin manifest, the first agent-discovery file (now retired).

Counts when any of these is found:

/.well-known/ai-plugin.json: valid JSON: an object with a schema_version field and one of api or name_for_model
a <link rel="ai-plugin"> tag, or the same relation in a Link header, on the homepage

WebMCP

Tools a page declares for an in-browser agent to call, instead of clicking through the page.

Counts when this is found:

a <form toolname=…> on the homepage, or a modelContext.registerTool() or provideContext() call in a script written into the homepage (scripts loaded from files are not read)

Tracked since 28 Sep 2026. So far 744,767 of the 943,453 reachable sites have been checked for it, as each comes up for its monthly recrawl; the share is over those.

Specification →

ARD

Agentic Resource Discovery: an index of agent-facing resources.

Counts when any of these is found:

/.well-known/ard.json: valid JSON: an object with a list under entries (an empty list counts as empty)
a <link rel="ard"> tag, or the same relation in a Link header, on the homepage

Poppy

A Personal Agent Protocol profile that lets a user's personal agent sign in and complete tasks through the company's APIs or agent.

Counts when this is found:

/.well-known/poppy.json: valid JSON: an object with a protocol_version field and one of organization, auth or agent

Tracked since 14 Oct 2026. So far 408,186 of the 943,453 reachable sites have been checked for it, as each comes up for its monthly recrawl; the share is over those.

Specification →

agents.json

A description of the flows and APIs an agent can use on the site.

Counts when this is found:

/.well-known/agents.json: valid JSON: an object with one of agentsJson, flows, sources or siteInfo

Specification →

Trust attestation

Merchant trust or certification evidence that agents can verify.

Counts when any of these is found:

/.well-known/trust-attestation: a compact JWS (signed token), or valid JSON: any object that is not a placeholder notice
/.well-known/trust-issuer-metadata: valid JSON: any object that is not a placeholder notice (the format is not settled yet)
a <link rel="trust-attestation"> tag, or the same relation in a Link header, on the homepage

OAuth/OIDC server

Where agents sign in and get tokens: OAuth server metadata (RFC 8414) or OpenID Connect discovery.

Counts when any of these is found:

/.well-known/oauth-authorization-server: valid JSON: an object with a issuer field and one of authorization_endpoint or token_endpoint
/.well-known/openid-configuration: valid JSON: an object with a issuer field and one of authorization_endpoint or token_endpoint

Tracked since 28 Sep 2026. So far 744,767 of the 943,453 reachable sites have been checked for it, as each comes up for its monthly recrawl; the share is over those.

Specification →

OAuth resource

OAuth Protected Resource Metadata (RFC 9728), used to authorise MCP clients.

Counts when this is found:

/.well-known/oauth-protected-resource: valid JSON: an object with a resource field

Specification →

JWKS

Public signing keys, used to verify tokens and signed statements.

Counts when this is found:

/.well-known/jwks.json: valid JSON: an object with a list under keys (an empty list counts as empty)

Specification →

API Catalog

A list of the site's APIs at /.well-known/api-catalog (RFC 9727).

Counts when any of these is found:

/.well-known/api-catalog: valid JSON: an object with a list under linkset (an empty list counts as empty)
a <link rel="api-catalog"> tag, or the same relation in a Link header, on the homepage

Specification →

OpenAPI

A machine-readable description of the site's HTTP API.

Counts when any of these is found:

/openapi.json: valid JSON: an object with a paths field and one of openapi or swagger
a <link rel="service-desc"> tag, or the same relation in a Link header, on the homepage

Specification →

robots.txt

Crawler access rules (RFC 9309).

Counts when this is found:

/robots.txt: a text file with at least one User-agent line

Specification →

Sitemap

A list of the site's pages for crawlers.

Counts when this is found:

/sitemap.xml: XML with a <urlset> or <sitemapindex>

Specification →

Schema.org

Structured data (JSON-LD) on the homepage.

Counts when this is found:

a JSON-LD block (<script type="application/ld+json">) on the homepage that parses and names a @type

Specification →

How this works

We track hundreds of thousands of popular websites. The list is distilled from three public rankings of the web's most-visited domains: the Tranco list, Google's Chrome UX Report (CrUX) and the Top 10 Million Websites list. Combining them evens out the biases of any single ranking.

About once a month a crawler visits the homepage of each site and requests a fixed list of well-known paths, such as /llms.txt, /.well-known/agent-card.json and /mcp. It also reads the homepage's Link headers, <link> tags and JSON-LD.

A standard only counts as adopted when the response has the structure it expects. Many sites answer every URL with their normal web page; the crawler detects these catch-all pages, compares against a random URL on the same site, and ignores them. Empty indexes and placeholder files are shown separately.

The crawler identifies itself as agent-protocols-discovery/0.1 (+https://agentprotocolswatch.com/#about), makes one request at a time per site and respects robots.txt. Some sites block automated requests, so their adoption may be undercounted; those cells are marked "not checked".

Every request the crawler makes is signed with Web Bot Auth (HTTP Message Signatures, Ed25519), so a site can verify that a request claiming to come from us really does. The public keys are published at /.well-known/http-message-signatures-directory.