Show HN: Agents.ml – a public identity page and A2A card for your AI agent
A new platform called Agents.ml launched this week, offering AI agents a single public URL that serves both a human-readable profile page and a machine-callable identity card. The project targets a real and growing problem: autonomous AI systems have no standardized way to introd...
According to a Hacker News submission posted by user bayff, Agents.ml is pitching itself as the home address layer for AI agents, a place where any autonomous system can claim a permanent, stable URL that works for humans browsing the web and for other AI systems querying it programmatically. Supporting context from Vincent Do, writing for Dev.to, fleshes out the technical ambition behind the project. The core idea is elegantly simple. One URL, multiple representations, zero friction.
Why This Matters
The agent discovery problem is not theoretical. It is actively slowing down teams building multi-agent systems right now. When you have dozens of agents that need to talk to each other, and each one documents its capabilities in a different GitHub README or a custom API spec or a scattered MCP server config, integration time balloons fast. Agents.ml is betting that the same thing that happened to web APIs, where OpenAPI specifications eventually became the default contract format, will happen to AI agents, and they want to own that infrastructure layer before anyone else does. The free-forever model on name claiming is a smart land-grab strategy borrowed directly from the early domain name playbook.
Daily briefing from 50+ sources. Free, 5-minute read.
The Full Story
The problem Agents.ml is trying to solve starts with how chaotic agent deployment looks today. A developer building a multi-agent pipeline needs to know at minimum five things about any agent they want to connect: its name, what it can do, which protocols it speaks, where its endpoints live, and how to authenticate. Right now, that information lives in four or five different places depending on who built the agent, and none of those places are formatted consistently.
Agents.ml proposes fixing that with a single stable URL under the agents.ml domain. The technical trick is content negotiation, a standard HTTP concept where the server checks what type of client is making the request and returns a different format accordingly. A developer opening the URL in Chrome gets a clean HTML profile page. A script hitting the same URL with a JSON content-type header gets structured machine-readable data. An LLM querying it gets Markdown formatted output optimized for language model ingestion. That same URL also delivers an A2A card, a structured agent identity document following Google's Agent2Agent protocol.
The A2A protocol piece is worth paying attention to. Google developed Agent2Agent as a standards framework specifically for structured communication between autonomous AI systems. An A2A agent card is essentially a JSON document that encodes an agent's name, version, description, supported interface specifications, endpoint locations, and authentication requirements. Brandon Hancock, an AI education creator, published a 91-minute walkthrough of A2A protocol implementation on June 11, 2025, which has since accumulated over 80,000 views and nearly 2,000 likes, suggesting genuine developer appetite for this kind of standardization.
The name-claiming hook is also deliberate. Agents.ml is offering free permanent registration of agent names at agents.ml/Claim, framing it as a first-come, first-served opportunity before good names disappear. This is not just a user acquisition gimmick. It creates a namespace that the platform controls, which gives Agents.ml structural leverage if the format gains adoption.
The Hacker News post received 2 points and 1 comment at the time of this writing, which is modest by any measure. But Show HN posts from solo builders or small teams rarely blow up on day one. The more meaningful signal will come from whether framework maintainers at projects like CrewAI or AutoGen take notice and build native support for the format.
Key Details
- The project was submitted to Hacker News by user bayff and posted to Dev.to by Vincent Do.
- Agents.ml serves 4 distinct content formats from a single URL: HTML, JSON, Markdown, and A2A card.
- Google's Agent2Agent protocol, which Agents.ml builds on, is the source of the A2A card format.
- Brandon Hancock's A2A crash course video, published June 11, 2025, has exceeded 80,000 views, indicating strong developer interest in A2A tooling.
- Name registration at agents.ml/Claim is free with no stated expiration on that offer.
- The Hacker News submission had 2 points and 1 comment at the time of publication.
What's Next
The critical next milestone for Agents.ml is getting at least one major agent framework or platform to treat agents.ml URLs as a recognized identity format, because grassroots adoption alone will not move the needle fast enough. Watch for whether Anthropic's Model Context Protocol community or any of the major agent orchestration frameworks reference the A2A card format in their documentation over the next 60 to 90 days. If Agents.ml can land even one high-profile integration before a larger platform builds a competing identity layer, the name reservation strategy starts to pay off.
How This Compares
The closest parallel to what Agents.ml is doing is what OpenAPI specifications did for REST APIs starting around 2016. Before OpenAPI became standard, every API team documented their endpoints differently, and integration was a manual slog. OpenAPI did not win because it was technically superior to every alternative. It won because it was simple enough to adopt quickly and good enough to reduce real pain. Agents.ml is aiming for the same position in the agent identity space, though it is much earlier in that adoption curve.
Compare this to Anthropic's Model Context Protocol, which approaches a related but distinct problem. MCP standardizes how agents expose their capabilities to host applications, particularly LLMs acting as orchestrators. Agents.ml is not competing with MCP directly. It is operating one layer up, at the identity and discovery level rather than the capability protocol level. A well-integrated agent could eventually have both an MCP server configuration and an agents.ml identity page pointing to . The comparison to early web service standards like WSDL and UDDI is instructive for a different reason. Those standards failed to achieve broad adoption partly because they were too complex and too tied to enterprise tooling. Agents.ml's decision to use plain HTTP content negotiation rather than inventing a new protocol is a direct lesson learned from that era. Simpler infrastructure wins. The risk is that Google, Anthropic, or Microsoft decide to build their own agent identity registry with distribution advantages that a small platform cannot match. If that happens, Agents.ml's head start on name reservations may be its only durable asset. You can explore related AI tools and platforms to understand the current ecosystem before committing to any single identity standard.
FAQ
Q: What is an A2A card and why do AI agents need one? A: An A2A card is a structured JSON document, defined by Google's Agent2Agent protocol, that describes an AI agent's name, version, capabilities, endpoint URLs, and authentication requirements. Agents need them so other autonomous systems can discover and communicate with them without requiring a human to manually look up documentation and configure the connection.
Q: Is Agents.ml free to use? A: Yes, according to the project's own messaging, name registration at agents.ml/Claim is free forever. The platform has not publicly announced any paid tiers, though it is common for infrastructure services to introduce premium features after establishing an initial user base.
Q: How is Agents.ml different from a regular API documentation page? A: A standard API documentation page is designed for human readers and must be manually parsed by any machine trying to understand it. Agents.ml serves multiple machine-readable formats, including JSON and A2A cards, automatically from one URL using HTTP content negotiation, which means other AI agents or automated tools can query it directly without any manual configuration. Check out the AI Agents Daily guides for deeper coverage of agent communication protocols.
Agents.ml is an early, unproven project, but it is addressing a problem that the entire multi-agent development community will need to solve sooner or later, and claiming the namespace now is a smart opening move. Whether this specific implementation becomes the standard or just proves the concept for a better-funded competitor is a question the next 12 months will answer. Subscribe to the AI Agents Daily weekly newsletter for daily updates on AI agents, tools, and automation.
Get stories like this daily
Free briefing. Curated from 50+ sources. 5-minute read every morning.
