Example: Node.js Project

A complete walkthrough of using Grove with a plain Node.js project — no Docker required.

Project structure

my-node-app/
├── package.json
├── src/
│   └── index.ts
└── .grove/
    └── config.json    # committed

package.json:

{
  "name": "my-node-app",
  "scripts": {
    "dev": "tsx watch src/index.ts",
    "start": "node dist/index.js"
  }
}

Setup

cd my-node-app
grove setup --preset node --yes

This writes .grove/config.json:

{
  "project": "my-node-app",
  "providers": {
    "web": { "type": "node-scripts" }
  },
  "naming": {
    "composeProject": "my-node-app-${branch_safe}",
    "ports": {
      "WEB_PORT": "auto"
    }
  },
  "worktrees": {
    "defaultBaseBranch": "main"
  }
}

Create a worktree

grove start feat/new-api --new

Creating worktree at ~/repos/my-node-app-worktrees/feat-new-api…
  branching off main
Generating .env.worktree from grove config…
  WEB_PORT=3001
Resolving environment provider…
  → provider: node-scripts
Starting environment…
  web: http://localhost:3001
  source: grove
✓ Ready: feat/new-api

Grove generates .env.worktree and starts npm run dev. The dev server picks up WEB_PORT if your script reads it, or Grove auto-detects the port from the process output.

Working across worktrees

# Two features running at once
grove start feat/new-api --new       # → http://localhost:3001
grove start feat/auth-rework --new   # → http://localhost:3002

# Open each in your editor
grove open feat/new-api
grove open feat/auth-rework

Monitor from the TUI

grove

The TUI shows both worktrees, their detected URLs, and their git change footprints.

Stop and clean up

grove stop feat/new-api       # stops the dev server
grove delete feat/new-api     # removes the worktree

Tip: port assignment

Grove allocates WEB_PORT from free ports starting above 3000. Your dev server reads this from .env.worktree. Make sure your start script reads process.env.PORT or process.env.WEB_PORT:

const port = process.env.WEB_PORT ?? process.env.PORT ?? 3000;