2. Your First Agent

With a working API provider in place, your first agent takes under five minutes.

Create a workspace

Click the workspace switcher in the top bar and choose Create Workspace. In the dialog:

1. Give it a name — my-first-agent, say
2. Choose a mode:
   • From Template — start from a template shared on your instance. A template carries a whole agent configuration (provider, model, prompt, sometimes skills). If one fits, take it and this step is done
   • Blank — configure it yourself. On a fresh install with no templates yet, this is the way, and it’s only a few fields:
     • Agent Type — Claude Code, Codex or Goose
     • API Provider — the list only shows providers that work with the type you picked (the mapping is in Getting Ready)
     • Model — one the provider actually serves
     • Prompt — leave it empty for now; write it once you’re inside

Click Create.

Go in

The workspace starts and opens on its own. The first start takes a few seconds while the platform brings up an instance for the agent.

It opens as three columns:

Area	What it’s for
Left — Session History	Switching between sessions, or starting one
Middle — working area	App tabs: Files / Browser / Skill Studio / Terminal / Automation / Memory / Settings — the agent’s environment and configuration
Right — Chat	Where you talk to it

Rearrange it however you like: move apps between columns, open more, or pop one into its own window. ⌘K reaches anything.

Open any configuration item and the right side of the dialog explains its fields. When something’s unclear, look there rather than coming back here.

Your first conversation

Type into the chat box on the right:

Hi, introduce yourself in one sentence

It replies. Follow up, send images, paste links, or give it something to do:

List the files in the current working directory

It runs the command in its own environment and comes back with the result.

With no prompt written yet, it answers as a generic assistant — able to talk, with no particular way of working. Come back after you’ve written the prompt and it’s a different agent.

A look at Files

Open the Files tab in the middle column and you’re looking at the agent’s working directory — the files it just listed for you. What you put there it can read; what it produces shows up there for you.

Files are shared across sessions. Start a new session and everything is still where it was.

Next

You have an agent that runs and holds a conversation. Making it do what you actually want comes next: a prompt for how it works, skills for its specialties, MCP to reach external systems, memory so it fits you better over time. That’s Defining Agent Behavior.