Agentic Resource Discovery

Every provider on the APIs.io network serving an AI Catalog manifest — at /.well-known/ard.json, or at the /.well-known/ai-catalog.json it replaced — the file that tells a discovery service what agentic resources a domain offers. Graded against the AI Catalog data model, then checked a second time with the ARD project’s own conformance tool, because the two do not always agree.

99 publishers · 78 pass the official tool · 472 advertised resources · 23 on the current path · 89,228 hosts probed

What is this? Agentic Resource Discovery (ARD) is a v0.9 draft proposal, published 28 May 2026 by authors at Microsoft, Google and Hugging Face with a contributors wall of eleven companies, for the layer that sits in front of every agentic protocol: the step before invocation where a client asks “what is available for this task?” Publishers describe their resources once in an AI Catalog manifest on their own domain; independent discovery services crawl those manifests and answer questions over them through POST /search. It does not execute anything — MCP, A2A and OpenAPI keep that job — and it deliberately refuses to be the one registry. This section holds both halves: who publishes a manifest, and what the registries actually return.

The finding, re-measured 2026-09-13. ARD v0.91 renamed the path: /.well-known/ai-catalog.json became /.well-known/ard.json, and rel="ai-catalog" became rel="ard". The old names are now a courtesy a consumer MAY consult and is not required to. Every publisher below was probed in the spec’s own resolution order — ard.json first, then the predecessor — and 76 of 99 answer only on the predecessor path. A consumer implementing only the current specification does not find them at all. apis.io serves the same body at both paths and emits both link relations. We grade our own manifest by the same rules as everyone else’s, and it is not conformant either.

The sweep, re-run 13 September 2026. Probing both paths across 89,228 hosts — every absolute host advertised in the catalog plus each host’s PSL-derived registrable apex, of which 81,109 answered — returned 175 hosts serving a manifest, which resolve to 119 distinct publishers. 99 of those map to a profiled provider and have a page below. The previous sweep, on 31 July 2026, probed 43,834 domains on the predecessor path alone and found 8. The two numbers are not comparable as growth: this run has a larger host list, probes a path the first one could not, and six weeks separate them.

The identity layer is decorative. did:web: is the affordance that separates a correctly shaped manifest from a conformant one, and 36 publishers declare one. One resolves. The rest return 404, 403 or 500 at the /.well-known/did.json their own identifier points at, and not one DID document in the corpus carries a verification method — so there is nothing to verify a signature against even where the document exists. An identifier that resolves to nothing is a string, not an identity, and this section no longer grades it as one: a declared-but-dead did:web: takes its own gap, did-web-does-not-resolve. That is why 0 publishers are conformant. Two were, for a few hours on 13 September, until the DID each one declared was actually fetched. The claim that two of the eleven organizations on ARD’s own contributors wall serve a manifest dates from the 31 July run and has not been re-measured.

Conformant

0 Every hard structural check in the data model passes, and the optional affordances — representativeQueries for semantic search, a did:web: host identifier that actually resolves, a wildcard CORS header, the spec’s own application/ai-catalog+json content type — are all present too.

None. The bar is not high — two to five natural-language queries per entry, a wildcard CORS header, the spec’s own application/ai-catalog+json content type, and a did:web: identifier that returns a DID document rather than a 404. That last one is what empties this band: 35 publishers would otherwise be here or close to it.

Near-conformant

66 Correctly shaped — an ARD client can parse these — but missing optional fields that cost the publisher search relevance or verifiability.

APIs.io

24 entries

APIs.io advertises 24 resources across 5 media types, with 51 representative queries for semantic search.

apis.io tool: PASS Kin Score 93.6

intlayer

18 entries

Intlayer advertises 18 resources across 5 media types, with 53 representative queries for semantic search.

intlayer.org tool: PASS

Apify

15 entries

Apify advertises 15 resources across 7 media types, with 18 representative queries for semantic search.

apify.com tool: PASS 15 trust manifests Kin Score 35.5

zapier

14 entries

Zapier advertises 14 resources across 7 media types, with 32 representative queries for semantic search.

zapier.com tool: FAIL

brand-api

12 entries

Brandfetch advertises 12 resources across 8 media types, with 39 representative queries for semantic search.

brandfetch.com tool: PASS

clickhouse

12 entries

ClickHouse advertises 12 resources across 4 media types, with 26 representative queries for semantic search.

clickhouse.com tool: PASS

surrealdb

10 entries

SurrealDB advertises 10 resources across 6 media types, with 26 representative queries for semantic search.

surrealdb.com tool: PASS

0xArchive

9 entries

0xArchive advertises 9 resources across 5 media types, with 30 representative queries for semantic search.

0xarchive.io tool: PASS Kin Score 71.7

the-ticket-fairy

8 entries

Ticket Fairy advertises 8 resources across 4 media types, with 25 representative queries for semantic search.

ticketfairy.com tool: PASS

coinapi

7 entries

CoinAPI advertises 7 resources across 4 media types, with 15 representative queries for semantic search.

coinapi.io tool: PASS

reducto

7 entries

Reducto advertises 7 resources across 3 media types, with 6 representative queries for semantic search.

reducto.ai tool: PASS

confident-ai

6 entries

Confident AI advertises 6 resources across 6 media types, with 18 representative queries for semantic search.

confident-ai.com tool: PASS

hookdeck

6 entries

Hookdeck advertises 6 resources across 3 media types, with 18 representative queries for semantic search.

hookdeck.com tool: PASS

padlet

6 entries

Padlet advertises 6 resources across 5 media types, with 17 representative queries for semantic search.

padlet.com tool: PASS

speakeasy

6 entries

Speakeasy advertises 6 resources across 6 media types, with 9 representative queries for semantic search.

speakeasy.com tool: PASS

Amplitude

5 entries

Amplitude advertises 5 resources across 3 media types, with 16 representative queries for semantic search.

amplitude.com tool: PASS Kin Score 56.3

calvis

5 entries

Calvis advertises 5 resources across 5 media types, with 19 representative queries for semantic search.

calvis.com tool: PASS

caretta

5 entries

Caretta advertises 5 resources across 4 media types, with no representative queries for semantic search.

caretta.so tool: FAIL 1 trust manifest

deepeval

5 entries

DeepEval advertises 5 resources across 4 media types, with 12 representative queries for semantic search.

deepeval.com tool: PASS

document360

5 entries

Document360 advertises 5 resources across 5 media types, with 11 representative queries for semantic search.

document360.com tool: PASS

duvo-ai

5 entries

Duvo advertises 5 resources across 4 media types, with 16 representative queries for semantic search.

duvo.ai tool: PASS

lago

5 entries

Lago Developer Documentation advertises 5 resources across 4 media types, with 14 representative queries for semantic search.

doc.getlago.com tool: PASS 5 trust manifests

ledger

5 entries

Ledger Developer Portal advertises 5 resources across 2 media types, with 18 representative queries for semantic search.

developers.ledger.com tool: PASS

quicknode

5 entries

Quicknode advertises 5 resources across 4 media types, with no representative queries for semantic search.

quicknode.com tool: PASS

supabase

5 entries

Supabase advertises 5 resources across 3 media types, with 12 representative queries for semantic search.

supabase.com tool: PASS

anew

4 entries

anew advertises 4 resources across 4 media types, with 13 representative queries for semantic search.

anew.page tool: PASS Kin Score 45.2

Aptos

4 entries

Aptos Developer Documentation advertises 4 resources across 4 media types, with 12 representative queries for semantic search.

aptos.dev tool: PASS Kin Score 49.5

bird

4 entries

Bird advertises 4 resources across 3 media types, with no representative queries for semantic search.

bird.com tool: PASS

chili-piper

4 entries

Chili Piper advertises 4 resources across 4 media types, with 12 representative queries for semantic search.

chilipiper.com tool: PASS

cloudflare

4 entries

Cloudflare Developer Documentation advertises 4 resources across 4 media types, with 8 representative queries for semantic search.

developers.cloudflare.com tool: PASS

fluentedi

4 entries

FluentEDI advertises 4 resources across 3 media types, with 10 representative queries for semantic search.

fluentedi.com tool: PASS

geico

4 entries

GEICO advertises 4 resources across 4 media types, with 12 representative queries for semantic search.

geico.com tool: PASS

leadping

4 entries

Leadping advertises 4 resources across 4 media types, with 12 representative queries for semantic search.

leadping.ai tool: PASS

limrun

4 entries

Limrun advertises 4 resources across 3 media types, with 12 representative queries for semantic search.

lim.run tool: PASS

listnr

4 entries

Listnr AI advertises 4 resources across 3 media types, with no representative queries for semantic search.

listnr.ai tool: PASS 4 trust manifests

lottiefiles

4 entries

LottieFiles advertises 4 resources across 4 media types, with 10 representative queries for semantic search.

docs.lottiefiles.com tool: PASS

pixeltable

4 entries

Pixeltable advertises 4 resources across 4 media types, with 6 representative queries for semantic search.

pixeltable.com tool: PASS

shorebird

4 entries

Shorebird advertises 4 resources across 3 media types, with 13 representative queries for semantic search.

docs.shorebird.dev tool: PASS

slng

4 entries

SLNG advertises 4 resources across 3 media types, with 12 representative queries for semantic search.

slng.ai tool: PASS

telnyx

4 entries

Telnyx advertises 4 resources across 2 media types, with no representative queries for semantic search.

telnyx.com tool: PASS 4 trust manifests

vercel

4 entries

Vercel advertises 4 resources across 2 media types, with no representative queries for semantic search.

vercel.com tool: PASS

anchor-x402

3 entries

anchor-x402 advertises 3 resources across 3 media types, with no representative queries for semantic search.

api.anchor-x402.com tool: PASS Kin Score 46.5

cal-com

3 entries

Cal.com advertises 3 resources across 3 media types, with 10 representative queries for semantic search.

cal.com tool: PASS

civitai

3 entries

Civitai Developer Documentation advertises 3 resources across 2 media types, with 6 representative queries for semantic search.

developer.civitai.com tool: PASS

flipdish

3 entries

Flipdish advertises 3 resources across 3 media types, with 9 representative queries for semantic search.

flipdish.com tool: PASS

fly-io

3 entries

Fly.io advertises 3 resources across 2 media types, with 9 representative queries for semantic search.

fly.io tool: PASS

helius

3 entries

Helius advertises 3 resources across 3 media types, with 9 representative queries for semantic search.

helius.dev tool: PASS

imper

3 entries

Imper.ai advertises 3 resources across 2 media types, with 6 representative queries for semantic search.

imper.ai tool: PASS 2 trust manifests

listen-labs

3 entries

Listen Labs advertises 3 resources across 3 media types, with 6 representative queries for semantic search.

listenlabs.ai tool: FAIL 3 trust manifests

mentimeter

3 entries

Mentimeter advertises 3 resources across 3 media types, with 9 representative queries for semantic search.

mentimeter.com tool: PASS

neuronto

3 entries

Neuronto ARD Registry: Agentic Resource Discovery (ARD) index advertises 3 resources across 3 media types, with 14 representative queries for semantic search.

neuronto.com tool: PASS 3 trust manifests

paz-ai

3 entries

Paz.ai advertises 3 resources across 3 media types, with 6 representative queries for semantic search.

paz.ai tool: PASS 3 trust manifests

roboflow

3 entries

Roboflow advertises 3 resources across 3 media types, with 15 representative queries for semantic search.

roboflow.com tool: FAIL

speko

3 entries

Speko advertises 3 resources across 3 media types, with no representative queries for semantic search.

speko.dev tool: PASS

thorchain

3 entries

THORChain advertises 3 resources across 3 media types, with no representative queries for semantic search.

thorchain.org tool: PASS

zuplo

3 entries

Zuplo advertises 3 resources across 2 media types, with no representative queries for semantic search.

zuplo.com tool: PASS 3 trust manifests

bizzabo

2 entries

Bizzabo advertises 2 resources across 2 media types, with 10 representative queries for semantic search.

bizzabo.com tool: PASS

brapi

2 entries

brapi.dev advertises 2 resources across 2 media types, with 5 representative queries for semantic search.

brapi.dev tool: PASS 2 trust manifests

datum

2 entries

Datum advertises 2 resources across 2 media types, with 7 representative queries for semantic search.

datum.net tool: PASS

hugging-face

2 entries

Hugging Face advertises 2 resources across 2 media types, with no representative queries for semantic search.

huggingface.co tool: PASS

josh-wood-colour

2 entries

Josh Wood Colour advertises 2 resources across 2 media types, with 8 representative queries for semantic search.

joshwoodcolour.com tool: PASS

loops

2 entries

Loops advertises 2 resources across 2 media types, with no representative queries for semantic search.

loops.so tool: PASS 2 trust manifests

xquik-api

2 entries

Xquik advertises 2 resources across 2 media types, with 4 representative queries for semantic search.

xquik.com tool: PASS 2 trust manifests

authologic

1 entry

Authologic advertises 1 resource across 1 media type, with no representative queries for semantic search.

developer.authologic.com tool: PASS

magichour

1 entry

Magic Hour advertises 1 resource across 1 media type, with 4 representative queries for semantic search.

magichour.ai tool: PASS

rudderstack

1 entry

RudderStack advertises 1 resource across 1 media type, with 5 representative queries for semantic search.

rudderstack.com tool: PASS

Flavored

33 A catalog in spirit rather than in schema: a hard check fails, so a client cannot assume the document parses to the model the specification defines. Often this is staleness rather than sloppiness — the urn:ai: publishers implemented against the identifier scheme ARD itself specified before ADR-0009 renamed the namespace to urn:air:.

infraspeak

15 entries

Infraspeak advertises 15 resources across 4 media types, with 48 representative queries for semantic search.

infraspeak.com tool: PASS

infisical

11 entries

infisical advertises 11 resources across 4 media types, with 31 representative queries for semantic search.

infisical.com tool: FAIL · 1

Airtable

10 entries

Airtable advertises 10 resources across 3 media types, with 29 representative queries for semantic search.

airtable.com tool: PASS Kin Score 61.5

datarobot

10 entries

DataRobot Agent Skills advertises 10 resources across 1 media type, with no representative queries for semantic search.

datarobot.com tool: FAIL · 2 pre-ADR-0009 urn:ai:

weaviate

10 entries

Weaviate advertises 10 resources across 5 media types, with 26 representative queries for semantic search.

weaviate.io tool: PASS

neon

9 entries

Neon advertises 9 resources across 2 media types, with no representative queries for semantic search.

neon.com tool: FAIL · 1

yasmina-holdings-limited

9 entries

yasmina-holdings-limited advertises 9 resources across 5 media types, with 23 representative queries for semantic search.

yasmina.ai tool: PASS

Alpic

6 entries

Alpic advertises 6 resources across 5 media types, with no representative queries for semantic search.

alpic.ai tool: FAIL · 1 Kin Score 48.6

gitbook

5 entries

GitBook advertises 5 resources across 3 media types, with 13 representative queries for semantic search.

gitbook.com tool: FAIL · 1

AI for Database

4 entries

AI for Database advertises 4 resources across 4 media types, with 6 representative queries for semantic search.

aifordatabase.com tool: PASS 4 trust manifests Kin Score 60.3

contextdev

4 entries

contextdev advertises 4 resources across 4 media types, with no representative queries for semantic search.

context.dev tool: PASS

explorium

4 entries

Explorium advertises 4 resources across 4 media types, with 3 representative queries for semantic search.

explorium.ai tool: PASS 4 trust manifests

hyperline

4 entries

Hyperline advertises 4 resources across 3 media types, with 10 representative queries for semantic search.

hyperline.co tool: PASS 1 trust manifest

mercury

4 entries

Mercury advertises 4 resources across 2 media types, with 8 representative queries for semantic search.

mercury.com tool: PASS

nylas

4 entries

Nylas CLI advertises 4 resources across 3 media types, with no representative queries for semantic search.

cli.nylas.com tool: FAIL · 1 pre-ADR-0009 urn:ai:

theta-edge

4 entries

Theta EdgeCloud advertises 4 resources across 4 media types, with 13 representative queries for semantic search.

www.thetaedgecloud.com tool: PASS

sendhq

3 entries

SendHQ advertises 3 resources across 3 media types, with 10 representative queries for semantic search.

sendhq.cc tool: PASS 3 trust manifests

Artifactories

2 entries

Artifactories advertises 2 resources across 2 media types, with 9 representative queries for semantic search.

artifactories.com tool: FAIL · 1 Kin Score 52.3

dupe

2 entries

Dupe.com advertises 2 resources across 2 media types, with 7 representative queries for semantic search.

dupe.com tool: PASS 2 trust manifests

guildai

2 entries

guildai advertises 2 resources across 2 media types, with 6 representative queries for semantic search.

docs.guild.ai tool: PASS

yuno

2 entries

yuno advertises 2 resources across 2 media types, with 6 representative queries for semantic search.

y.uno tool: FAIL · 2 pre-ADR-0009 urn:ai:

Aave

1 entry

Aave advertises 1 resource across 1 media type, with no representative queries for semantic search.

aave.com tool: PASS Kin Score 15.8

Appwrite

1 entry

Appwrite advertises 1 resource across 1 media type, with no representative queries for semantic search.

appwrite.io tool: FAIL · 2 Kin Score 82.3

Artie

1 entry

Artie advertises 1 resource across 1 media type, with no representative queries for semantic search.

artie.com tool: FAIL · 2 Kin Score 48.0

bitso

1 entry

Bitso advertises 1 resource across 1 media type, with 3 representative queries for semantic search.

bitso.com tool: PASS

c1

1 entry

c1 advertises 1 resource across 1 media type, with no representative queries for semantic search.

c1.ai tool: FAIL · 2

conductorone

1 entry

conductorone advertises 1 resource across 1 media type, with no representative queries for semantic search.

conductorone.com tool: FAIL · 2

domscan

1 entry

domscan advertises 1 resource across 1 media type, with no representative queries for semantic search.

domscan.net tool: FAIL · 2

fundraiseup

1 entry

Fundraise Up advertises 1 resource across 1 media type, with no representative queries for semantic search.

fundraiseup.com tool: FAIL · 1

honeybook

1 entry

HoneyBook advertises 1 resource across 1 media type, with 4 representative queries for semantic search.

honeybook.com tool: PASS

opus

1 entry

OpusClip advertises 1 resource across 1 media type, with 10 representative queries for semantic search.

opus.pro tool: FAIL · 1 pre-ADR-0009 urn:ai: 1 trust manifest

routerplex

1 entry

routerplex advertises 1 resource across 1 media type, with no representative queries for semantic search.

routerplex.com tool: FAIL · 2

xata

1 entry

xata advertises 1 resource across 1 media type, with no representative queries for semantic search.

xata.io tool: FAIL · 2

The discovery services

4 The four reference implementations ARD names, exercised over the endpoints each one documents. Index size is the registry’s own count, read from POST /explore facets where it implements them.

Registry	Operator	Index size	POST /search	GET /agents envelope	Referrals
Hugging Face Discover `https://huggingface-hf-discover.hf.space`	Hugging Face	no /explore	4/5 task queries answered	detail	1
GitHub Agent Finder `https://agentfinder.github.com/api/v1`	GitHub	1188	5/5 task queries answered	offset, pageSize, results, total	0
Ora Directory `https://ora.ai/api/ard`	Ora	5518	5/5 task queries answered	items[]	1
the operator's edge resets connections from a non-browser client; identical requests succeed from curl, so every Ora figure here was collected that way
Cisco AI Catalog (AGNTCY) `https://ai-catalog.outshift.io/v1`	Cisco	no /explore	no results	nextPageToken, results, totalCount	0

Federation is specified, implemented, and empty. Federation is specified, implemented and empty. Across the four reference registries the only referral returned by anyone points at a second registry run by the same operator; no registry refers a client to another vendor. The specification’s whole premise is many independent registries referring clients onward to one another; today the only referral anyone returns points at a second index run by the same operator.

Method. Every absolute Website, Portal, MCPServer, Documentation and APIReference host in the catalog, plus each host’s derived apex, was fetched at /.well-known/ai-catalog.json, the path current at the time — 43,834 domains, 37,751 of them reachable. A manifest is graded, never counted: a boolean “publishes a catalog” would report every publisher here as an ARD adopter, and most of them fail a hard check. Each body is graded against the AI Catalog 1.0 data model and run through the ARD project’s own conformance CLI v0.9.1 — against the captured body rather than over the wire, because three of these publishers’ edges return 403 to that tool’s HTTP client, which is a problem for crawlers too. Attribution is owner-only: a manifest on a shared host belongs to the host’s owner, never to every provider that links there. The verbatim JSON for each publisher is committed to that provider’s repository in the api-evangelist organization, so any grade here can be checked against the bytes it was computed from. Nothing on these pages is derived or generated — a manifest is served from the publisher’s own domain or it does not exist. Read the full analysis in The ARD Standard.