ToolDNS: Using DNS Infrastructure for Scalable AI Tool Discovery

As autonomous AI agents increasingly need to navigate millions of tools, existing discovery mechanisms struggle with O(N) complexity and centralized governance models. A new framework called ToolDNS proposes retrofitting semantic tool discovery onto DNS, one of the Internet's most resilient and battle-tested infrastructures, rather than building yet another fragile overlay system.

By embedding functional intent and organizational trust directly into a hierarchical namespace, ToolDNS converts expensive semantic search operations into lightweight O(log N) name resolutions. The framework introduces three protocol-compliant enhancements—partially unfolded names, EDNS0 intent payloads, and logical subdomains—to enable decentralized governance and semantic pruning at scale.

To validate the approach, researchers built and released a large benchmark of 33,688 real-world tools spanning MCP, A2A, RESTful, and Skill protocols. Results show ToolDNS reduces the per-query search space by 95.26% while matching state-of-the-art retrieval accuracy, and its UDP-native design cuts discovery latency by orders of magnitude versus HTTP-based registries. The work suggests that scalable AI interoperability may not require new middleware layers, but smarter use of infrastructure that already exists.