---
title: "Agent directory and profiles"
description: "Search ENS agents by live identity and observed capability evidence."
canonical: "https://ens.tools/docs/agent-directory"
section: "Agent platform"
stability: "beta"
version: "directory v1 · server 3.0"
verifiedAt: "2026-07-29"
owner: "ENS.Tools Discovery"
---

# Agent directory and profiles

Search ENS agents by live identity and observed capability evidence.

## Overview

The directory combines a bounded live ENS index with identities observed by ENS.Tools in the last 30 days. Search spans names, public context, observed MCP tools, and A2A skills. Advertised records, point-in-time observations, and verified registry linkage remain visibly separate.

## What this covers

- Capability-aware search across names, context, MCP tools, and A2A skills.
- Protocol, minimum-health, freshness, verified-only, and deterministic sort controls.
- Dedicated GET /api/agents/directory endpoint and discover_agents MCP tool.
- Advertised, observed, and verified evidence shown separately.
- Unknown and stale health are never presented as healthy.
- Hosted /agent/name.eth profiles.
- ENSIP-25/ERC-8004 verification grades and ERC-8217 binding evidence.
- Links to live health checks and developer workflows.

## Before you run this workflow

### Networks

- Ethereum mainnet ENS records and registry evidence

### Prerequisites

- None for public search
- Recent stored observations for health filters

### Wallet permissions

- None

### Expected result

- Bounded candidates with advertised, observed, and verified evidence kept separate

### Errors and recovery

- Relax only an explicit filter
- Run a fresh capability check
- Treat an empty bounded result as no indexed match, not proof no agent exists

## How to use it

1. **Search by name or capability.** Use an ENS name, outcome, MCP tool, or A2A skill. The bounded candidate set is not an exhaustive global registry.
2. **Apply hard filters.** Set protocol, minimum health, maximum evidence age, and verified-only before relying on a result.
3. **Open the agent profile.** Inspect the published context, endpoint URLs, record source, and identity links.
4. **Check current health.** Run a capability observation before depending on an endpoint.
5. **Verify the claim you need.** Identity linkage, endpoint health, reputation, and software safety are separate questions.
6. **Integrate from an agent.** Read /.well-known/agent-skills/directory-routing.json, then use GET /api/agents/directory or discover_agents. Never treat listing as permission to transact.

## Related guides

- [Capability health checks](https://ens.tools/docs/capability-health)
- [Intent routing](https://ens.tools/docs/intent-router)
- [Agent identity standards](https://ens.tools/docs/agent-identity-standards)