Skip to main content
Ready to put your first AI squad member to work? This guide walks you through deploying a squad member using Desktop — from opening the wizard to chatting with your new teammate.

​
Prerequisites

Before you begin, make sure you have:
  • Desktop running — the desktop app (npm run mc:dev). See Getting Started for setup instructions.
  • At least one AI provider connected — your squad member needs an API key for the model it will use. If you haven’t set one up yet, head to AI Providers to configure one.

​
Deploy step by step

1

Open the Deploy wizard

In Desktop, click Squad in the sidebar, then Deploy. This opens the deployment wizard where you’ll configure your new squad member.
2

Choose a name

Give your squad member a name. This is how you’ll identify it in the squad list, in chat, and in logs. Pick something descriptive — like “code-reviewer” or “research-assistant”.
3

Select a model

Choose which AI model your squad member will use. Dash supports models from Anthropic (Claude), OpenAI (GPT), and Google (Gemini). If you’re not sure, anthropic/claude-sonnet-4-20250514 is a great default — it balances speed and capability.
4

Write a system prompt

The system prompt tells your squad member who it is and how to behave. Be specific about what you want it to do. For example:
A clear system prompt makes a big difference in how well your squad member performs.
5

Pick tools

Select which tools your squad member can use. Tools let your squad member interact with the outside world — reading files, running shell commands, searching directories, and more. See Tools for the full list.Common starter sets:
  • General assistant — bash, read, write
  • Code-focused — bash, read, write, edit, find, grep
  • Read-only — read, find, grep
6

Add fallback models (optional)

You can configure fallback models in case your primary model is unavailable. Fallbacks are tried in order — if the first model fails, Dash tries the next one. This is optional but useful for reliability.
7

Deploy

Click Deploy. If you left the workspace blank, Desktop creates one for you under ~/.dash/workspaces/<agent-id>/ — you can open this folder in Finder or your editor to drop files in for the squad member to work on. The squad member process starts and is added to the squad list, and you’ll see a confirmation with the deployment details.
8

Verify it's running

Open the squad member detail page by clicking on your new squad member in the Squad list. Check that:
  • Status shows “running”
  • The Monitor tab shows uptime and the squad member’s process ID (PID)
If the status shows “error”, check the logs tab for details.
9

Chat with it

Click Chat to open a conversation with your new squad member. Type a message and hit Enter. You’ll see the squad member thinking, using tools, and responding in real time.Congratulations — your first squad member is deployed and ready to work!

​
What’s next

Squad member configuration

Learn more about squad member settings, models, and advanced configuration.

Connect Telegram

Let your agent respond to messages on Telegram or WhatsApp.

Add skills

Give your agent reusable instructions for specific tasks.