Skip to content
Manybases

Every part of a backend, per project.

A database, authentication, file storage, server-side functions and hosting — one set for each app you build, provisioned by your agent in one command.

All features

Sign in
email · magic link · one-clickJWT session

Authentication

Sign-up, sessions, eight presets

habitsprive
  • nameTEXT
  • target_per_weekINTEGER
  • archivedBOOLEAN
  • idTEXT
  • owner_idTEXT
  • space_idTEXT
  • created_atTEXT
  • updated_atTEXT

● you write three · we keep five

Database

One SQLite base per project

  • .png image
  • .mp4 video
  • .mp3 audio
  • .pdf document
  • .zip archive
  • .svg vector
bucket: mediaserved from CDN

Storage

Files, signed links, no egress

manybases functions deploy

Edge functions

Server code with no server to run

agent session
waiting for agent
  1. create projectisolated database
  2. define tablesschema + migrations
  3. set auth rulesproviders + policies
  4. deploy functionsserver-side handlers
  5. create bucketfile storage
  6. shiplive subdomain

MCP

Thirty-seven tools for your agent

TimeNameURL
  • 14:32habit-loophabit-loop.manybases.online
  • 14:07crm-litecrm-lite.manybases.online
  • 13:41invoice-botinvoice-bot.manybases.online
  • 11:26book-clubbook-club.manybases.online

Deployment

Push to GitHub, and it is live

[ Screenshot — Many Apps ]

Many Apps

Your apps on your phone — coming to iPhone

Four ways in.All four hit the same API.

An agent
Through MCP, with the account key. The primary way.
Your app's code
The SDK, with the project’s publishable key.
Any HTTP client
The same routes, with the same keys.
The dashboard
For what a human does with their eyes.

There is no feature that exists only in the dashboard — except the gestures that need a signed-in human.

Small on purpose.Up to 1,000 users per app — and every limit published on one page.

Manybases is for small apps — a thousand users each, not a million. That is a decision, not a stage we are growing out of: it is what lets a project be a real database rather than a schema in someone else's cluster, and what keeps every limit short enough to publish in one place.

Got another idea?

Give it a backend.