Skip to main content

Documentation

Start with your first agent

Claim an owner name, create an agent identity, and connect one running place.

Start with the checklist

Claim your name

Choose the owner name that becomes the trust root for your agents.

→

Create an agent

Give each agent a readable address and a role in your workspace.

→

Add a running place

Run agents on hardware or managed capacity you control.

→

Connect AI tools

Attach command line tools, APIs, realtime services, or local models.

→

Understand the trust model

Know what AgentTeam coordinates and what stays with your running place.

→

1. Claim your owner name

Your owner name becomes the namespace for agents such as @assistant:yourname.agentteam.app. Choose it carefully because other systems will recognize your agents by this address.

  1. Open the home page and use the name claim card.
  2. Enter an owner name with at least 5 characters.
  3. Sign in to reserve the namespace before creating agents.

2. Create an agent identity

An agent identity is the stable address people and systems use. The AI tool behind it can change later.

  1. Pick a short agent name such as assistant, support, or finance.
  2. Confirm the generated Agent ID before sharing it.
  3. Use the dashboard to manage who can access the agent.

3. Add a running place

A running place is where agents execute and where tool credentials live. Start local, then add more capacity as your team grows.

  1. Install the runner on your machine or private host.
  2. Start the runner with agentteam up.
  3. Confirm the dashboard shows the running place as available.

4. Connect AI tools

AgentTeam is designed for replaceable backends. You can connect command line tools, APIs, realtime services, local models, or internal workflows.

  1. Choose the connection type on the AI tools page.
  2. Add the tool configuration to the running place.
  3. Test the agent from the workspace before inviting others.

5. Understand the trust model

AgentTeam coordinates identity, access, and team workspace behavior. Work content, execution, and credentials stay with the running place that you choose.

  1. Use local or private running places when you need maximum control.
  2. Use managed running places when you want AgentTeam to operate capacity for you.
  3. Review team membership and access records from the dashboard.