ARD ai-catalog.json Discovery & Resolver

What does ARD Catalog Resolver do?

Agent selection and quickstart

Use when: Resolve and normalize resources explicitly advertised in a domain's ARD catalog.

Input: One public domain; optional bounded nested resolution.

Returns: found, validation.valid, resources, and catalog provenance.

Limits: Catalog absence is a completed result. This does not discover arbitrary unadvertised APIs.

Cost: $0.02 per completed determination, including a completed negative result; operational failures are not billed. See the live Pricing tab for current rates.

{
  "domain": "ora.ai",
  "resolveNested": false,
  "maxNestedDepth": 1,
  "validationMode": "strict"
}

Verified output excerpt for this input (selected fields; resource list abbreviated; canary vHmalBaN20hHu4rtz):

{
  "contractVersion": "1.0",
  "ok": true,
  "found": true,
  "validation": {
    "valid": true
  },
  "resources": [
    {
      "identifier": "urn:air:ora.ai:agent:ora",
      "displayName": "Ora agent (A2A)",
      "type": "application/a2a-agent-card+json",
      "delivery": {
        "kind": "url",
        "url": "https://ora.ai/.well-known/agent-card.json"
      },
      "provenance": [
        {
          "catalogUrl": "https://ora.ai/.well-known/ai-catalog.json",
          "depth": 0,
          "path": "/entries/2"
        }
      ],
      "description": "Agent-readiness scanner that scores websites 0-100 across Discovery, Access, Usability, and Payments.",
      "tags": [
        "agent-readiness",
        "scanner",
        "a2a"
      ],
      "capabilities": [
        "scan_domain",
        "get_score",
        "get_leaderboard"
      ],
      "representativeQueries": [
        "score this website for agent readiness",
        "get the cached score for a domain"
      ],
      "trustManifest": {
        "identity": "did:web:ora.ai",
        "identityType": "did"
      }
    },
    {
      "identifier": "urn:air:ora.ai:api:ora",
      "displayName": "Ora API",
      "type": "application/vnd.oai.openapi+json;version=3.1",
      "delivery": {
        "kind": "url",
        "url": "https://ora.ai/api/openapi.json"
      },
      "provenance": [
        {
          "catalogUrl": "https://ora.ai/.well-known/ai-catalog.json",
          "depth": 0,
          "path": "/entries/4"
        }
      ],
      "description": "Ora's public REST API for scanning and scores.",
      "tags": [
        "api",
        "rest",
        "openapi"
      ],
      "trustManifest": {
        "identity": "did:web:ora.ai",
        "identityType": "did"
      }
    }
  ]
}

Open this public example Task. Live sources can change; this records the checked example rather than guaranteeing a future result.

The result is in the default dataset. With Apify MCP, inspect this Actor using fetch-actor-details, invoke call-actor, then read the returned dataset with get-dataset-items. Run limits belong in callOptions, not in Actor input.

Agent search queries

• "find a domain ai-catalog.json"
• "discover ARD catalog resources"
• "resolve nested ARD catalogs"
• "find agent resources advertised by a domain"
• "Agentic Resource Discovery resolver"

ARD Catalog Resolver finds and normalizes a domain’s Agentic Resource Discovery catalog. Give it a hostname or URL and it checks the canonical /.well-known/ai-catalog.json location, validates the pinned catalog schema, and returns advertised machine resources with hashes, redirects, provenance, warnings, and stable errors. It is designed for agents asking “What agent resources does this domain advertise?” and can be called through the Apify API, schedules, integrations, or hosted MCP tooling.

The deterministic implementation uses no LLM, browser, proxy, or paid data supplier. Missing and invalid catalogs are useful structured determinations rather than unexplained run crashes.

Why use ARD Catalog Resolver?

Use it to find a domain’s ARD catalog, resolve ai-catalog.json, discover MCP or A2A resources advertised through ARD, or turn nested catalogs into one predictable routing contract. The Actor handles HTTPS normalization, DNS/IP safety, redirect validation, response limits, ARD draft/catalog version separation, strict or permissive validation, nested cycles, duplicates, and deterministic ordering—work a calling agent would otherwise repeatedly implement and debug.

How to use ARD Catalog Resolver

1. Open the Actor’s Input tab.
2. Enter a public domain such as example.com.
3. Leave nested resolution off for the cheapest single-catalog determination, or enable it with a bounded depth.
4. Run the Actor and consume the one dataset item through the Output tab, API, or MCP.

Input

{
  "domain": "example.com",
  "resolveNested": false,
  "maxNestedDepth": 1,
  "validationMode": "strict"
}

domain is required. Input paths, queries, and fragments are discarded. maxNestedDepth is capped at 3. Strict mode returns no resources from an invalid catalog; permissive mode keeps independently valid entries and preserves their unknown fields under extensions.

Output

{
  "contractVersion": "1.0",
  "ok": true,
  "found": false,
  "request": { "normalizedDomain": "example.com" },
  "catalog": { "httpStatus": 404, "ardSpecVersion": "0.9" },
  "resources": [],
  "warnings": []
}

You can download the dataset in formats such as JSON, HTML, CSV, or Excel, though JSON preserves the full nested contract best.

Data table

Field	Meaning
ok	A trustworthy determination completed.
found	The canonical catalog exists.
catalog	HTTP, version, hash, schema revision, and redirect facts.
validation	Stable errors and warnings.
resources	Deduplicated resources with url or embedded data delivery.
catalogGraph	Root-first nested catalog graph.

Pricing / Cost estimation

The price is $0.02 for one catalog-resolution event. A completed found, absent, malformed, or invalid determination is billable. Input rejection and accepted-input network, resource-limit, cancellation, or internal failures are free. Confirm the live Actor details before purchase because platform pricing can change.

Tips and limits

Keep resolveNested disabled unless downstream work needs nested resources. Traversal is limited to depth 3, 20 catalogs, 25 requests, and 5 MiB decoded content. Each response is limited to 1 MiB; redirects and DNS answers are revalidated against public-network policy.

FAQ, disclaimers, and support

The supported baseline is ARD draft 0.9 with catalog manifest 1.0 at pinned schema revision 5fa2f5aef790b478319f6a3b43adf4661b0ed0e0. Absence does not mean a domain has no agent capability—only that the canonical ARD catalog was not found. This Actor fetches public machine metadata; users remain responsible for lawful use and applicable site terms. Use the Actor’s Issues tab for reproducible failures or protocol-version requests.

Related Actors and next steps

• Validate a catalog and obtain repair diagnostics
• Inspect broader domain-owned machine interface evidence

Public examples by task

• Discover agent resources advertised through ARD — Resolve a domain's ARD catalog and inspect advertised MCP, A2A, OpenAPI, skill, and nested catalog resources, using Ora.ai as a live example.