The Linux Foundation launched DNS-AID, an open-source project that uses the internet’s existing DNS infrastructure as a decentralized directory for AI agents. Here’s why this matters for the future of agentic AI.
AI agents are popping up everywhere. They book your meetings, write your code, analyze your data, and talk to each other. But here’s the dirty little secret of the agentic AI boom: agents are terrible at finding each other.
Right now, if you build an AI agent and want it to connect with another agent built by a different team or company, you have two options. Hardcode the URL, or register both agents on a centralized platform that acts as a middleman. Neither scales. Neither is secure. And neither feels like the open internet we all signed up for.
The Linux Foundation just fixed that. On May 27, 2026, it announced DNS-AID (DNS for AI Discovery) – an open-source project that gives AI agents their own phonebook built directly on top of the Domain Name System (DNS). No new infrastructure. No proprietary registries. Just DNS, but for bots.
In This Article
What Exactly Is DNS-AID?
DNS-AID is an open-source naming convention and protocol that lets AI agents publish, discover, and verify each other using standard DNS records. Think of it as the yellow pages for AI agents – but way more secure and running on infrastructure that’s already been battle-tested at planetary scale for over four decades.
Initially developed by Infoblox, the project is now hosted under the Linux Foundation with backing from Cloudflare, Equinix, GoDaddy, CSC, Indeed, Internet Systems Consortium (ISC), and WWT. It uses existing DNS record types – SVCB (Service Binding), TXT, and TLSA – so there are zero new protocols to deploy.
Ingmar Van Glabbeek, project maintainer for DNS-AID, put it simply: “Current approaches to agent connectivity are fragmented and often rely on fragile, hardcoded configurations. With DNS-AID, we are moving toward a ‘web-native’ model for AI.”
Why AI Agents Need a Discovery Layer
The agentic web is growing fast. McKinsey estimates that agent-to-agent commerce could represent a $3 trillion to $5 trillion economic opportunity in the coming years. But none of that happens if agents can’t find each other.
Here is the problem in plain English:
- Hardcoded URLs break – When an agent moves or updates, every agent that pointed to it breaks too.
- Centralized registries become chokepoints – Whoever runs the registry controls who gets to play. That is the opposite of decentralization.
- No standard format – Every platform invents its own way to describe what agents do, what protocols they speak, and how to trust them.
DNS-AID solves all three. It treats agent discovery the same way the internet treats website discovery – using a distributed, hierarchical, cached, and globally understood system that just works.
How DNS-AID Works (Without the Jargon)
DNS-AID piggybacks on something every developer already knows: DNS. When you type google.com into your browser, DNS translates it into an IP address so your computer knows where to go. DNS-AID does the same thing, but for AI agents.
The Four-Step Agent Discovery Flow
- Publish – A developer publishes an SVCB record for their agent under their domain’s _agents subzone. This record includes the agent’s endpoint, protocol (MCP, A2A, HTTPS), and capabilities.
- Sign – The domain’s authoritative DNS server signs the records with DNSSEC, creating a cryptographic chain of trust from the root of the internet down to that specific agent.
- Discover – A remote agent (or a human) queries DNS for the SVCB record by agent name, capability type, or full domain index.
- Connect – The discovering agent validates DNSSEC + DANE signatures, then connects directly to the agent’s endpoint using whatever protocol the record describes.
- MCP (Model Context Protocol) – Anthropic’s standard for connecting models to tools
- A2A (Agent-to-Agent) – Google’s emerging protocol for agent interoperability
- HTTPS – Standard web APIs
- Anything else addressable via SVCB and ALPN
- AWS Route 53, Azure DNS, Google Cloud DNS – all major cloud DNS providers
- Cloudflare – edge infrastructure and security
- Infoblox – the original developer of DNS-AID
- Equinix – global interconnection and data center leader
- GoDaddy – the world’s largest domain registrar
- Internet Systems Consortium – the team behind BIND9, the most widely used DNS server
- Install the SDK:
pip install dns-aid-core - Initialize:
dns-aid init - Publish an agent:
dns-aid publish --name my-agent --protocol mcp --endpoint https://my-agent.example.com/mcp - Discover agents on any domain:
dns-aid discover example.com - Verify trust:
dns-aid verify _my-agent._mcp._agents.example.com - Agent marketplaces without a middleman – Anyone can publish an agent under their own domain and be discovered globally
- Cross-enterprise agent workflows – Company A’s planning agent can discover and talk to Company B’s scheduling agent without a shared platform
- Trust-by-design – DNSSEC verification means you can prove an agent is who it claims to be, using the same cryptographic chain that secures banking and government sites
- No vendor lock-in – DNS-AID works with any DNS provider, any cloud, and any agent protocol
- Use-case fit: who this is for and who should skip it.
- Practical value: what changes for developers, creators, teams, or businesses.
- Trust check: claims are compared against public product pages, announcements, docs, and observable market context when available.
Here is the beautiful part: the naming format is completely human-readable. An agent record might look like _my-chatbot._mcp._agents.example.com. You can query it with a standard dig command:
dig SVCB _chatbot._mcp._agents.example.com
No special client. No proprietary API. Just a DNS query.
What Protocols Does DNS-AID Support?
DNS-AID is protocol-agnostic. It works with all the major agent communication protocols today:
This is a big deal. Developers aren’t locked into a single vendor’s ecosystem. DNS-AID doesn’t care which framework you use – it just gives your agent an address on the global agentic web.
Who Is Backing DNS-AID and Why It Matters
The list of initial supporters reads like a who-is-who of internet infrastructure. Cloudflare CTO Dane Knecht said it best: “The Internet already solved the discovery problem decades ago with DNS – it’s fast, it scales globally, and every network on earth understands it.”
Other supporters include:
When the people who literally run the internet’s DNS infrastructure get behind a project, you pay attention. This isn’t a garage startup experiment. DNS-AID has the operational muscle to actually ship and sustain itself at internet scale.
How Developers Can Get Started Today
DNS-AID is open source and ready to use right now. The project provides a Python SDK on GitHub, a command-line interface, and an MCP server.
Getting started takes minutes:
For local testing, you can spin up a Docker-based BIND9 playground that requires no credentials. The project even includes a testing docker-compose file to get you running in seconds.
The Bigger Picture: DNS-AID and the Future of Agentic AI
DNS-AID is more than just a technical standard. It is a philosophical statement about how the agentic web should work. The internet succeeded because it was built on open, decentralized protocols that no single company controlled. DNS-AID takes the same approach for AI agents.
Think about what this unlocks:
Dane Knecht of Cloudflare called DNS-AID “the foundational routing layer that autonomous systems need to operate safely and efficiently.” He is right. Without a discovery layer, the agentic web is just a collection of walled gardens. With DNS-AID, it becomes an extension of the open internet.
For more AI tools and technology reviews, check out AIToolGate.com – your guide to the best AI tools and the latest AI news.
Looking for more AI coverage? Browse our latest articles for hands-on reviews of AI tools, deep dives into emerging AI technologies, and analysis of the trends shaping the AI landscape. Bookmark AIToolGate.com and stay ahead of the AI curve.
How I reviewed this
AI Tool Gate evaluates AI tools and AI industry updates from a developer/operator perspective. I look at practical use cases, product positioning, pricing signals, reliability concerns, and whether the tool is actually useful for real workflows.
Written by
Gallih Armadaw
Senior backend developer with 8+ years of experience building production systems across PHP/Laravel, Node.js, cloud infrastructure, Web3, and AI-assisted workflows. I review AI tools from a practical developer/operator perspective.