AI APP backend, ready out of the box.

Every AI APP needs the same backend chores — data storage, user auth, file uploads, realtime sync, and serverless compute. Stop reinventing them.

Start free

Open source under Apache 2.0 · Self-host or run on Mortar Cloud

Why Mortar

6 cost dimensions, 1 bill

3 baseline (compute / database / cache — always-on capacity) + 3 metered (storage / network / function — pay per use) — shaped like your cloud bill

SDK in 4 languages

TypeScript · Swift · Kotlin · Dart. AI agents already know the API shape (Supabase-style)

Open core

Apache 2.0 — every primitive, every feature, every SDK. Mortar Cloud is the managed wrapper, not a fork

Tier sizing

Nano / Micro / Small / Medium / Large — start free on Nano, size up by capacity. Switch sizes without re-architecting

Quick start · three ways in

SDK · inside your app code

npm install @mortar/client

import { createClient } from '@mortar/client';
const mortar = createClient({
  url: 'https://api.mortar.appunvs.com',
  apiKey: 'mtr_live_...',
  tenant: 'proj_abc',
});

const todos = await mortar.from('todos').select({ limit: 10 });
await mortar.from('todos').insert({ title: 'buy milk' });

CLI · terminal / CI / agent container

# One-line login + provision; same commands in CI / scripts / agent containers
npx @mortar/cli login --new
mortar projects create --name=my-app --tier=tiny
mortar keys create admin --scope=admin --save

# Subsequent commands target the current project
mortar usage me
mortar tables list
mortar compute deploy worker ./worker.js

MCP · let AI agents call Mortar directly

// Drop into Claude Desktop / Cursor / Codex CLI config:
{
  "mcpServers": {
    "mortar": {
      "command": "npx",
      "args": ["-y", "@mortar/mcp"]
    }
  }
}

// Tools the AI gets, with no extra glue:
//   mortar_query           read rows
//   mortar_insert          write one row
//   mortar_usage_me        credit + tier-cap snapshot
//   mortar_projects_list   all projects on the account