Setup

Run moona setup inside the project you want to guard. One command does everything — you will be protected before it finishes.

Run setup

What setup does

moona setup is the single onboarding command. In order, it:

1. Checks your Node version. Fails immediately if you are running Node older than 20.

2. Initializes configuration if this is a new project. Creates a moona.config.json in the current directory. Any undeclared database host is treated as production by default — the safe starting point.

3. Starts the guardrail daemon (or reuses one already running).

4. Detects and configures your agent integrations. Moona checks which of the four supported agents — Claude Code, Cursor, VS Code, Codex — you already have configured and wires each one automatically. Existing configuration files are merged, never overwritten.

5. Discovers databases by reading your .env files, Docker Compose configuration, and Prisma schema. This is read-only: Moona never opens a database connection or reads your passwords during this step. Any remote host it finds is automatically treated as production.

6. Prints a summary and next steps.

Non-interactive environments

By default, moona setup is fully automatic — it makes sensible decisions without prompting. Two flags give you explicit control in CI or scripted environments:

Re-discovering databases

When your project gains a new database connection — a new .env entry, a new service in Docker Compose — run moona scan to bring it into Moona's view:

moona scan re-reads your project's database configuration (read-only, never opening a connection) and lets you declare each discovered host as production or non-production. Remote hosts default to production. Skipping a host is always safe — undeclared hosts are already treated as production.