From zero to running your first AI agent. No prior experience needed.
Claude Code executor: squads-cli uses Claude Code to run agents locally. Make sure it's installed and authenticated with your Anthropic API key before proceeding.
$ npm install -g squads-cli
Verify the install: squads --version
Navigate to your project directory (or create a new one), then run:
$ squads init
The -e flag runs the agent immediately (non-interactive). This is the fastest way to see results.
$ squads run engineering -a backend-engineer -e What just happened: squads-cli read your SQUAD.md goals, loaded the backend-engineer agent definition, and ran it via Claude Code. The agent acted on your goals and saved output to memory.
Edit .agents/squads/engineering/SQUAD.md and add a real goal for your project. Then run the agent again — it reads your goals every time.
## Goals - Write unit tests for the auth module - Review and document the API endpoints
$ squads run engineering -a backend-engineer -e The example squad has 3 agents. Run any of them, or add your own. Here's where to go next: