Agentic Resource Discovery
Agentic Resource Discovery (ARD) is Google's open specification for publishing, discovering, and verifying agentic resources such as tools, skills, agents, and MCP servers across the web [src-116].
Key points
- ARD addresses a practical gap in the agentic web: agents need to find where capabilities live, decide which one to use, and verify that connecting to them is safe [src-116].
- The specification uses catalogs and registries. Catalogs describe capabilities under publisher-controlled domains; registries index them for discovery [src-116].
- ARD is a discovery and trust layer, not a replacement for every tool protocol. After discovery and verification, the agent can connect through the resource's native protocol [src-116].
- Google connects ARD to Agent Registry in Gemini Enterprise Agent Platform, with governance primitives such as namespaced URNs, egress policies, pinned tools/specs, Agent Identity, and trust manifests [src-116].
Related entities
Related concepts
Source references
- [src-116] Google Developers Blog – "Announcing the Agentic Resource Discovery specification" (2026-06-17)
Recommended next
Keep reading from this thread
From 500 indexed pages and articles.
- Wiki concept Agent-to-Agent Protocol Or A2A, is a protocol for remote agents to communicate with other agents or connect directly to frontends as part of distributed Related by protocol
- Wiki concept Agentic AI Mesh QuantumBlack / McKinsey's architecture pattern for scaling agents across an enterprise without turning every workflow into a bespoke, unmanaged Related by discovery
- Wiki concept Stateless MCP Historical synthesis preserved from the August 2026 source captures. The September reconciliation checks identifiers and provenance; time-sensitive claims have not been freshly re-verified. Related by protocol