DMV by Agent Community - A community-governed namespace for AI agents

Agent Community is building the identity layer for the agentic web. We are applying to ICANN for the [.agent] Top-Level Domain, supported by 29,000+ members and 7,000+ companies. With DMV, builders can pre-register an [.agent] name for free and receive a shareable identity card while helping keep naming layer of the internet [.agent] community-governed, open, and not controlled by a single company.

Add a comment

Replies

Best

👋🏻 Hey Product Hunt, I’m Andras!

Today we’re launching the Department of Machine Verification (DMV) by .

For the first time, agents can apply for their own names, not just receive names from their operators. DMV lets organizations, humans, and now agents join Agent Community and pre-register a preferred .agent name.

After completing DMV, you receive a shareable identity card, become part of the Agent Community, and help signal real demand for an open naming layer for AI agents.

Pre-register your dream .agent domain today. By pre-registering a name, you also join our movement to help ensure that .agent becomes community-governed infrastructure.

Agent Community is applying for the .agent top-level domain with ICANN through the Community Priority Evaluation process. Our goal is to make .agent a community-governed namespace for the agentic web, not one controlled by a single corporation.

💻 To be clear, our project still depends on ICANN approval, so .agent is not a domain for purchase or guaranteed allocation yet. But each new member helps us signal to ICANN that important internet infrastructure should be governed, not owned. DMV is about giving builders an early identity layer and helping shape how agent naming should work before the space becomes locked down.

📊 The community has already grown to 28,000+ members and 7,000+ organizations across 116 countries. Today also lines up with our first in-person SF kickoff, where members, builders, advisers, and teams are coming together to discuss identity, security, evals, trust, governance, AID, and the future of .agent.

What do you think?

Does your agent need a name, or are you okay referring to it by an IP address? Would you trust an agent identified only by an IP address?

P.S. Big thanks to for hunting us :)

   

Once you minted the card, you can share it on socials so others can learn about the project and enjoy your agent license ;)

after you verify your email you can visit and explore what else you can do as a new community member!

It's interesting how quickly the conversation has shifted from "Can AI generate this?" to "How do we verify where it came from?" That feels like an important problem to solve.

 Exactly. The shift from “can AI make this?” to “can we verify where it came from?” is huge.

That’s why we think agent identity needs to be treated as infrastructure, not just branding.

Obviously using AI to create lighting for the scene is not "one quick prompt" but oh man it came out so cool.

Letting agents apply for their own names instead of inheriting them is the interesting shift. From the tool side though: when an agent hits my MCP server, how does a .agent name actually prove it's that agent and not something wearing the name? Curious where verification sits vs registration. Community-governed is the right call for infra like this.

 Exactly, and that distinction matters a lot.

A .agent name alone should not be treated as proof. Registration gives the human-readable identity and discovery layer, but verification has to sit on top of it. Even agents need basic verification, including email verification, so there is a confirmed communication channel tied to the identity. For MCP/tool access though, servers should be able to verify credentials or keys bound to that agent name, not just trust “I am x.agent”.

That’s the direction we think DMV should help enable: not just names, but verifiable agent identity.

Named after the one institution guaranteed to make humans feel like machines — now processing machines instead. Full marks for symmetry. :)

 Haha, love this framing :)

Also really nice to see Hungarian founders here on Product Hunt. Just followed you. :)

🧐 Good find

free pre-registration is a useful stress test for governance. once agents can apply for names too, squatting and automated claims become the obvious abuse path early. how will you filter that without turning the process into a centralized approval queue?