With MCP and A2A, agents now have protocols for accessing tools, analyzing data, and collaborating with one another. But one crucial step had yet to be addressed with a standardized solution: how do you find the right resource, select it from among several options, and verify that it is trustworthy before using it?
That is precisely the goal of Agentic Resource Discovery (ARD). This open specification aims to catalog agents, MCP servers, APIs, and workflows, and then make them accessible through registries that can be queried using natural language. It could thus facilitate the creation of internal directories, collaboration between organizations, and the governance of resources used by agents.
Still in the draft stage, ARD is not ready for widespread adoption without careful consideration. But the support of several major technology players makes it a development worth watching closely—and perhaps a future fundamental infrastructure for agent-based AI.
How does ARD actually work, how does it complement MCP and A2A, and what new opportunities does it open up for businesses?








