Agent Discovery

Agent discovery in ACP helps you quickly find available agents and understand their capabilities. This simplifies integration and ensures compatibility. Agent metadata is standardized through the Agent Detail specification and can be extended to support your custom needs.

​

Basic Discovery (via ACP Server)

You can discover agents hosted on ACP servers using REST APIs or the ACP Python SDK. Basic discovery returns essential info like:

• Name and description
• Capabilities
• Operational status

curl http://localhost:8000/agents

curl http://localhost:8000/agents

import asyncio
from acp_sdk.client import Client

async def main():
    client = Client(base_url="http://localhost:8000")
    async for agent in client.agents():
        print(agent.name, agent.description)

asyncio.run(main())

​

Registry-Based Discovery

Registry-based discovery provides a centralized view of agents across multiple ACP servers. This makes managing and discovering agents easier, especially in large-scale deployments or enterprise environments.

It enables:

• Centralized agent listings
• Scalable management across multiple servers
• Simplified search and discovery

While not yet part of the official ACP spec, this feature is already implemented in the BeeAI Platform.

We are actively gauging community interest in making this a core part of the ACP specification. To help shape the direction, visit our Contribute page.

​

Offline Discovery

Offline discovery allows agents to be discoverable without requiring network connectivity. This is achieved by embedding agent metadata directly into the distribution package, typically through container image labels or bundled metadata files.

By embedding metadata during build time, the Agent Detail remains synchronized with the agent’s actual implementation. This approach:

• Prevents mismatches between advertised capabilities and actual behavior
• Provides value in secure, disconnected, or automated deployment environments
• Simplifies distribution and deployment workflows

We recommend embedding Agent Detail metadata directly into container images during the build process. The BeeAI Platform follows this best practice by storing standardized metadata in container image labels, which:

• Streamlines integration with CI/CD pipelines
• Ensures metadata consistency throughout the deployment lifecycle
• Simplifies agent discovery in offline environments

​

Open Discovery (Manifest Files)

Open discovery allows Agent Detail to be made publicly accessible through a standardized manifest file. This makes it easy for apps, websites, and third-party tools to discover agent capabilities by knowing only the agent’s domain.

We recommended that you publish agent metadata using a YAML file (agent.yml), placed in a well-known public location:

https://your-domain.com/.well-known/agent.yml

The Agent Detail specification currently covers only the description of an agent’s capabilities. It does not include instructions for how to deploy or consume the agent.

• Managed agents (packaged for deployment) require standardized build and deployment instructions
• Unmanaged agents (already running services) require metadata on how to access and consume the service

These deployment details are not yet part of the official ACP specification. However, we are exploring options for standardizing both managed and unmanaged agent metadata. Early prototypes have used containers, uvx, and npx to simplify agent distribution and usage.

To participate in ongoing discussions about manifest standardization and best practices, visit our Contribute page or join these existing discussions:

• Agent Manifest File Format #427
• Manifest-Based Agent Offline Discoverability #344