What is the Agentic Resource Discovery (ARD) specification?

ARD is an open standard, not a product, released under the Apache License 2.0. It defines a common protocol and format for describing agentic resources, enabling federation across registries. AWS contributed feedback during its development.

How does the AWS Agent Registry workflow operate?

An administrator creates a registry with authorization via IAM or JWT. Publishers submit records for approval, a curator approves or rejects them, and consumers search the registry to discover approved resources.

What makes AWS Agent Registry enterprise-ready?

It offers an approval workflow for curation, hybrid search with semantic understanding and keyword matching, MCP-native access for any compatible client, and flexible authorization using IAM or corporate JWTs.

Open-Source AIAI Business & IndustryAmazon AI BlogAug 24, 2026Aug 25, 2026, 04:00 JST

AWS launches Agent Registry and ARD open spec for cross-environment AI agent discovery

Send on LINE

3 Key Points

What happened
AWS announced AWS Agent Registry, a centralized catalog for AI agents, MCP servers, tools, and skills, now in preview. It also contributed to the Agentic Resource Discovery (ARD) specification, an open standard under Apache License 2.0 for cross-environment discovery.
Why it matters
Without a central catalog, agents stay siloed across clouds, on-premises, and SaaS. ARD enables federation across registries without bilateral agreements, analogous to how DNS enables name resolution, so resources described once can be discovered everywhere.
What to watch
ARD is available at agenticresourcediscovery.org and on GitHub, with reference implementations. AWS expects ARD to enable cross-environment discovery while keeping local control, and to open cross-organizational discovery paths for Agent Registry customers.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

This launch addresses the growing challenge of discovering AI agents and tools as organizations scale their use of them. Previously, developers had to manually locate and connect resources across different environments, a process that does not scale with the increasing number of agents and MCP servers. The introduction of ARD offers a common protocol to describe and discover these resources across clouds, on-premises, and SaaS platforms, reducing the need for bespoke connectors.

AWS Agent Registry serves as a centralized, searchable catalog within AWS, featuring curation, hybrid search, and MCP-native access. ARD complements this by acting as an interoperability layer, allowing organizations to federate without migrating resources, which could enable public discovery paths where any organization can publish a catalog on its own domain. The expectation is that this will open new cross-organizational discovery pathways while preserving local access controls.

FAQ

What is the Agentic Resource Discovery (ARD) specification?: ARD is an open standard, not a product, released under the Apache License 2.0. It defines a common protocol and format for describing agentic resources, enabling federation across registries. AWS contributed feedback during its development.
How does the AWS Agent Registry workflow operate?: An administrator creates a registry with authorization via IAM or JWT. Publishers submit records for approval, a curator approves or rejects them, and consumers search the registry to discover approved resources.
What makes AWS Agent Registry enterprise-ready?: It offers an approval workflow for curation, hybrid search with semantic understanding and keyword matching, MCP-native access for any compatible client, and flexible authorization using IAM or corporate JWTs.

Amazon AI BlogRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.