Agentic Resource Discovery (ARD) Engine Build

The protocol layer that lets AI agents find the business and verify who runs it. Quantum implements the Agentic Resource Discovery v1.0 spec (the Google/Cisco/Hugging-Face-sponsored standard) on the client’s domain — publishing a signed ai-catalog.json and a did:web:{domain} identity document at /.well-known/, advertising both through all four required channels (well-known file, robots.txt Agentmap directive, <link rel> head element, HTTP Link header), and adding the WAF allowlist that keeps managed-rule updates from blocking agent crawlers.

This is the “business license + sign on the door” for the agent web. Our OKF engine build tells agents what the business knows; ARD tells agents that this business exists, what services it offers, and who legally runs it. The two are complementary — and as of June 2026, ARD is the only published interop standard with multi-vendor sponsor backing.

Deliverables (one-time build + ongoing)

Build (one-time):

• ai-catalog.json authored, schema-validated against the official ARD JSON Schema, published at /.well-known/ai-catalog.json with correct MIME type

• did.json published at /.well-known/did.json binding did:web:{domain} to the site identity, signed off by site owner

• All four discovery channels live and verified (well-known, robots.txt Agentmap, HTML <link rel="ai-catalog">, HTTP Link: header)

• WAF / bot-rule allowlist installed for /.well-known/* (Cloudflare Skip rule, AWS WAF, or nginx — verified across seven UA probes including Python-urllib)

• Catalog + DID document committed to client’s source control

• Section 10 verification suite added as post-deploy CI step

• Full handoff packet: verification suite output, schema validation log, change-control runbook