_agent Identity & Discovery
DNS for agents. Discover endpoints and verify identity.
One _agent TXT record for instant discovery and agent identity you can verify.
Quick Start
Discover, publish and validate in minutes
import { discover } from '@agentcommunity/aid'
const { record } = await discover('example.com')
console.log(record.uri) // https://api.example.com/mcpThe Agent/Tooling Integration Problem
Connecting to AI agents shouldn't require a PhD in API archaeology
Manual Integration Hell
Each agent needs bespoke code, doc digging and manual config.
Protocol Fragmentation
Agents speak MCP, A2A, OpenAPI and more — auth flows vary wildly.
Wasted Development Time
Teams lose weeks wiring basic discovery and connection logic.
No Discovery & Identity Standard
No universal way to discover agents and verify who runs them.
How Agent Discovery Solves This
Three simple principles that eliminate integration complexity
One DNS TXT Record
Add a single _agent.example.com TXT record. That's it. No registries, no APIs, no complexity.
Instant Discovery
Any tool can find your agent by looking up the _agent subdomain. Falls back to .well-known/agent when DNS is restricted.
Protocol-Agnostic
Works with any agent protocol — just change the p= token in your record.
Agent Identity
Publish a public key (PKA) and let clients verify your endpoint using HTTP Message Signatures (Ed25519).
Identity (PKA)
Public Key for Agents: verify that you are connecting to the right endpoint.
How it works
PKA adds a public key (`k`) and a key id (`i`) to your `_agent` TXT record. Clients send a small challenge and verify an HTTP signature from your server using that key.
In short: DNS tells clients where to connect; Identity (PKA) lets clients verify who is on the other end. All keys have single-letter aliases for byte efficiency (e.g. u for uri, p for proto).
- Publish `k` (public key) and `i` (key id) in your TXT record
- Client sends an `AID-Challenge` to your `uri`
- Server returns an HTTP signature (Ed25519) covering the request
- Client verifies the signature using `k`
_agent.example.com. 300 IN TXT "v=aid1;\
uri=https://api.example.com/mcp;\
p=mcp;\
k=z7rW8rTq8o4mM6vVf7w1k3m4uQn9p2YxCAbcDeFgHiJ;\
i=g1"Complete Developer Toolkit
Everything you need to build AID-powered applications
Core Engine
Pure business logic for discovery, validation, identity
CLI – AID Doctor
CLI wrapper around aid-engine: validate & generate records
Conformance Suite
Golden fixtures and CLI runner for parity checks
Web Workbench
Try AID in the browser – no install
Coming soon
Open a PR
TypeScript / JS
SDK for Node.js & browser
Go
High-performance Go client
Python
Idiomatic Python client
Rust
Idiomatic Rust client
Java
Idiomatic Java client
.NET
C#/.NET client
The Future of AI Integration
Building the infrastructure for the next generation of AI applications
Universal Agent Ecosystem
Imagine a world where every AI service is instantly discoverable. No more hunting through documentation, no more custom integration code. Just type a domain and connect to any agent, anywhere.
- Cross-platform interoperability
- Automatic capability discovery
- Zero-config networking
- Verifiable agent identity (PKA)
Open-Source Agent Infrastructure
We're building a vendor-neutral stack for hosting, scaling and observing agents. Curious? Get involved at agentcommunity.org.
- MIT-licensed core
- Self-host or cloud
- Token registries & open index
- Community governance