Quickstart

Get up and running with ChooChoo in four steps. Make sure you have installed the CLI first.

1. Initialize a project

Run choochoo init inside any repository. It scaffolds a choochoo/docs content directory with example files for domains, teams, data products, data contracts, schemas, and more.

choochoo init

You can pass --dir ./path/to/docs to change the output location, or --yes to skip prompts.

See Project Structure for a full breakdown of the generated directories.

2. Validate

Run the validator against the scaffolded examples. ChooChoo auto-detects file types (ODPS, ODCS, OpenLineage, AI System Card) from file content.

choochoo validate ./choochoo/docs

You should see a summary of files checked and their validation status. If any files have errors, the CLI prints the file path, field path, and what went wrong. See Error Messages for details.

You can also validate individual files or use --schema to force a specific schema type:

choochoo validate contract.yaml --schema odcs

3. Browse your docs

ChooChoo includes two ways to explore what you just created:

Terminal UI — a keyboard-driven interface for browsing traces, docs, and schemas:

choochoo tui

Web UI — a local Fumadocs documentation server:

choochoo web

Both commands accept --dir to point at a custom content directory.

4. Connect your coding agent

ChooChoo can auto-detect coding agents (Claude Code, OpenCode, Gemini CLI) on your machine and install trace hooks so every AI-assisted change gets recorded.

# See what agents are installed
choochoo trace hooks scan

# Install trace hooks into their config files
choochoo trace hooks install

Use --dry-run to preview changes before writing. See Agent Traces for more on how tracing works.

Next steps