Got an existing A2A agent that is not wrapped using Agent Stack SDK? No problem, you can still connect it to Agent Stack by providing a Docker image or its running URL.Documentation Index
Fetch the complete documentation index at: https://agentstack.beeai.dev/llms.txt
Use this file to discover all available pages before exploring further.
Add Docker A2A Agent
When you want your agent to be managed via Agent Stack, provide a Docker image:Requirements
Your A2A agent Docker image must:- Expose a valid agent card at
/.well-known/agent-card.json - Start without required environment variables — Agent Stack temporarily runs the container to extract metadata, then terminates it. Your image must be able to start with sensible defaults (no mandatory env vars).
How It Works
When you register an A2A agent:- Agent Stack pulls and starts your Docker image
- Probes
/.well-known/agent-card.jsonto extract agent metadata - Terminates the container
- Uses the metadata to make the agent available in the interface
Add Unmanaged A2A Agent
If your A2A agent is already running and you don’t want Agent Stack to manage its lifecycle (unmanaged), you can add it directly via its URL:Requirements
Your running A2A agent must expose a valid agent card at/.well-known/agent-card.json. For the example above, this would be at http://localhost:8080/.well-known/agent-card.json.
How It Works
When you register an unmanaged A2A agent, Agent Stack fetches the agent card from the/.well-known/agent-card.json endpoint of the provided URL. It then enhances the metadata from the card to make the agent available in the interface.
Agent Stack will not manage the lifecycle of this agent. When users interact with it, requests are proxied to the provided agent URL.