TeamOnly docs

Quick start

You need Node.js 20 or newer and a TeamOnly account. From your project folder:

$npx teamonly init✔Added the TeamOnly skill, Cursor rules and SDK>Build an expense approval tool(your AI builds the app)$npx teamonly deploy✔Live at https://expenses.acme.teamonly.appOnly @acme.com accounts can sign in

The first deploy signs you in, creates the app, its database and its file storage. Later deploys update the same app at the same address.

CLI commands

npx teamonly login

Sign in to TeamOnly from this computer. Opens a browser.

npx teamonly init

Add the skill file (Claude Code), rules file (Cursor) and SDK to the current project.

npx teamonly deploy

Build the app and put it online. Creates the database and file storage on the first deploy. Prints the app’s address.

npx teamonly logs

Stream the app’s live logs, including errors.

npx teamonly secrets set NAME

Store a secret for this app. You are prompted for the value, so it never lands in your shell history.

SDK

Your API is a Hono app. The SDK gives each request a context with four things:

ctx.user

The signed-in person: email and name. Verified by TeamOnly before your code runs. Never read identity from headers yourself.

ctx.db

This app’s own SQL database (SQLite, on Cloudflare D1). Created on first deploy.

ctx.files

This app’s file storage: put, get, list, delete.

ctx.secrets

Values set with npx teamonly secrets set. Read with ctx.secrets.get(name).

import { Hono } from 'hono'
import { teamonly } from 'teamonly/sdk'

const app = new Hono()

// Who is signed in. Verified before your code runs.
app.get('/api/me', (c) => {
  const ctx = teamonly(c)
  return c.json(ctx.user) // { email: 'dana@acme.com', name: 'Dana Okafor' }
})

// Save to this app's own database.
app.post('/api/expenses', async (c) => {
  const ctx = teamonly(c)
  const { amount, reason } = await c.req.json()
  await ctx.db
    .prepare('INSERT INTO expenses (amount, reason, submitted_by) VALUES (?, ?, ?)')
    .bind(amount, reason, ctx.user.email)
    .run()
  return c.json({ ok: true })
})

// Store an uploaded receipt in this app's file storage.
app.put('/api/receipts/:id', async (c) => {
  const ctx = teamonly(c)
  await ctx.files.put(`receipts/${c.req.param('id')}`, c.req.raw.body)
  return c.json({ ok: true })
})

// Read a secret set with: npx teamonly secrets set QUICKBOOKS_KEY
app.get('/api/sync', async (c) => {
  const key = teamonly(c).secrets.get('QUICKBOOKS_KEY')
  // ...
})

export default app

The skill file

init adds a skill file for Claude Code and the same content as a rules file for Cursor. Your AI tool reads it before writing code. It teaches:

• The one supported app shape (below), and to stay inside it.
• To use ctx.user for identity and never write its own login, sign-up or password screens.
• To use ctx.db and ctx.files instead of adding a database service, Express, Supabase or Firebase.
• To keep API keys in ctx.secrets, never in code.
• How to run deploy, read logs and fix what they show.

The point is that the first deploy works. If your AI keeps reaching for something we don’t run, tell us; that is a bug in the skill file.

Supported app shape

• Front end: React, built with Vite, served as static files.
• API: a Hono app in TypeScript or JavaScript, under /api, on the same address.
• Runtime: Cloudflare Workers. Standard web APIs; most Node built-ins through compatibility, but no native modules or long-running processes.
• Not supported yet: Python, Ruby, PHP, Docker containers, background jobs, scheduled tasks, sending email, custom domains, public apps.

Limits

Final limits will be published here before launch. Today’s plan:

• Builder: 1 builder, 3 apps. Team: unlimited builders and apps.
• Each app’s database and file storage follow Cloudflare D1 and R2 limits.
• Requests run under Cloudflare Workers limits for CPU time and memory.

More on how it all fits together: How it works and Security.