Quickstart

Get from zero to your first AI-assisted edit in a couple of minutes.

1. Install

curl -fsSL https://assets.cluster.app/serve/cstatic-assets/releases/cli/install.sh | sh

This installs the cluster binary to a user-writable directory on your PATH (typically ~/.local/bin) so it can keep itself up to date; set INSTALL_DIR to choose another location. Confirm it's on your PATH:

cluster --version

2. Log in

cluster login

This signs you in with your Cluster account. On a desktop machine it opens your browser; on an SSH/remote shell — or a Linux/BSD session with no graphical display — it automatically uses the device flow instead, and it also falls back to the device flow if the browser can't be opened.

Force a specific flow when the automatic choice is wrong:

cluster login --device    # device-code flow for headless/remote machines
cluster login --browser   # browser flow with a localhost callback

Check who you're signed in as with cluster whoami.

3. Start chatting

From inside your project, run cluster with no arguments to open an interactive session:

cluster

Describe what you want done — "find where rate limiting is configured and add a 5-second timeout" — and the assistant works in your current directory: reading files, running commands, and proposing edits. When it wants to change a file, it asks for approval first. Press Ctrl-C to stop a turn in progress, and type /quit to exit.

4. Run a one-shot prompt

For scripts and quick questions, cluster exec runs a single prompt without entering the interactive UI:

cluster exec "summarize what this service does"

It reads from stdin too, so you can pipe context in:

echo "fix the failing test in api_test.go" | cluster exec

Next steps

• Interactive sessions — tools, approvals, plan mode, keybindings, and the full slash-command reference.
• Exec & sessions — scripting and resuming saved sessions.
• Models & configuration — switch models and tune settings.