Getting Started

The shortest path from zero to a working clu setup is three steps:

1. Install the binary.
2. Run clu init inside a project directory.
3. Create issues, claim work, close them.

There is no account to set up, no service to start, and no network access at any point. clu writes to a single file at .clu/data.sqlite and reads its configuration from .clu/config.yaml.

When you're ready to coordinate more than one agent, continue to multi-agent setup. When you want to capture repeatable sequences with human approval gates, see workflows.