Google published a new open specification called Agentic Resource Discovery (ARD), aimed at a problem nobody has actually standardized yet: how does an agent find a capability it doesn't already know about, and how does it know that capability is what it claims to be? The blog post frames it directly — agents need reliable answers to three questions: Where does the right capability live? Which capability should I actually use? And how do I verify it's safe to connect to? Today there's no shared answer to any of them across organizational boundaries. The mechanism is built on two primitives. Catalogs are files an organization publishes describing what it offers — MCP servers, A2A agents, OpenAPI tools, even nested catalogs — hosted at a well-known path under the org's own domain. Registries then crawl and index those catalogs, acting as search engines for the agentic web, and when an agent queries one, it gets back matching capabilities plus the metadata needed to verify the publisher before connecting. What's notable is where trust anchors. Google states that because these catalogs are hosted directly under the organization's own domain, ownership of that domain serves as the cryptographic foundation for identity and trust. That's a specific design choice: rather than inventing a new identity layer or relying on a central authority, ARD reuses DNS ownership as the root of trust, then layers cryptographic verification metadata on top so a client can confirm a publisher's identity before it ever opens a connection. The stated failure mode this is meant to fix is fragmentation — plenty of platforms already run their own capability registries, but they're siloed. Google describes existing setups as fragmented and siloed within specific ecosystems, preventing agents from easily communicating across different tools. ARD's pitch is to sit above all of them as a shared discovery and verification layer, then get out of the way once a connection is established, handing off to whatever native protocol — MCP, A2A, plain OpenAPI — the resource actually speaks. Google says the spec was built with unnamed industry partners, and the reference implementation is hosted publicly on GitHub under the ards-project organization. It's early: a spec repo and a demo video, not yet a deployed federation of registries. But it's a concrete attempt to answer the discovery question that's been open since agents started needing to find each other across domains rather than within a single vendor's walled garden — and it does so by betting on domain ownership, not a new PKI, as the trust root.