Updated August 2026

Migrate Your Supabase Project via Chat

One prompt. One API key. Your entire Supabase backend — schema, data, edge functions, auth — migrated while you watch. Works with Lovable, Bolt, Firebase, and Base44.

Published April 25, 2026

•Updated August 3, 2026

•5 min read

Staticbot's MCP server exposes 44+ tools that AI agents can call directly. That means you can run a full Supabase migration from your terminal — just tell your AI agent what to migrate and where. No dashboard, no clicking, no copy-pasting credentials between tabs.

Since launch, we've added support for Bolt, Firebase, and Base44 migrations, self-hosted Supabase targets with downloadable packages, live preview deployments to verify your app works before switching over, and a Chrome extension bridge that makes Lovable sync fully hands-free.

What You Need

An MCP-compatible AI agent

Claude Code, Cursor, Windsurf, or any tool that supports the Model Context Protocol

A Staticbot API key

Generate one in your Staticbot dashboard under Developer → API Keys

Your project details

GitHub repo URL, source platform (Lovable/Bolt/Firebase/Base44), and target Supabase project

Supported Platforms

Staticbot's migration pipeline adapts to your source platform automatically:

Lovable + Supabase

Full 8-phase pipeline with Chrome extension bridge for hands-free edge function deployment.

Bolt + Supabase

Same pipeline tuned for Bolt.new-built apps on Supabase.

Firebase

6-phase pipeline with AI-assisted Firestore-to-Postgres schema design. Requires Firebase service account JSON.

Base44 + Supabase

Base44 apps backed by Supabase. Backend switchover updates Base44 platform secrets.

Base44 Native

Base44 apps using the @base44/sdk against Base44's managed backend. DDL synthesized from entity schemas.

Step 1: Connect the MCP Server

Add the Staticbot MCP server to your AI agent's configuration. The server is published as staticbot-mcp on npm — no install needed, npx downloads and runs it automatically. For Claude Code, add this to your .mcp.json:

{
  "mcpServers": {
    "staticbot": {
      "command": "npx",
      "args": ["-y", "staticbot-mcp"],
      "env": {
        "STATICBOT_API_KEY": "sk-your-api-key-here",
        "STATICBOT_API_URL": "https://app.staticbot.dev"
      }
    }
  }
}

For Cursor, Windsurf, or other MCP clients, point them at npx -y staticbot-mcp with the same environment variables. See the MCP Server docs for full setup instructions.

Step 2: Paste the Prompt

Open your AI agent's chat and describe what you want. The agent will use the MCP tools to gather what it needs — reading your .env, finding integrations, and creating the migration. Here are some examples:

Lovable / Bolt

Migrate my Lovable project to my own Supabase.

My repo is https://github.com/username/my-project
My Lovable project ID is abc123-def456

Target Supabase project ref: xyz789

Self-hosted Supabase (downloadable package)

Migrate my Bolt project to my own self-hosted Supabase instance.

I want the downloadable package so I can apply it myself.

That's it. Your AI agent will use the Staticbot MCP tools to orchestrate the entire migration — finding integrations, parsing credentials, presenting the discovery inventory, and walking you through each decision gate.

What Happens Next

The AI agent calls Staticbot's MCP tools step by step. Here's the full pipeline — and where it'll ask for your input:

Setup

Auto

Reads your .env to find the source Supabase URL and anon key. Lists connected integrations (Supabase, GitHub). Creates or selects a template from your repo.

Discovery

Auto

Inventories your source project — tables, edge functions, storage buckets, auth providers, migration files. Presents the inventory for your explicit approval.

DB Migration

Auto

Applies all SQL migration files and enables required extensions on the target.

Data Import

Auto

Deploys an export edge function, calls it to export your data, and imports it to the target. Also copies storage objects, secrets, cron jobs, and OAuth identities.

Edge Functions

Auto

Deploys all your edge functions to the target Supabase project.

Auth Config

Auto

Updates auth configuration — providers, redirect URLs, site URL.

Backend Switchover

Auto

Asks how you want to handle the switch — fully switch to the new Supabase, keep preview on the old one, or park the migration as a backup.

Preview & Hosting

Auto

Builds a live preview of your app on Staticbot's infrastructure so you can verify everything works. Then lets you choose Staticbot hosting or host elsewhere.

Continuous Sync

Auto

Sets up automatic GitHub-to-target sync so every push keeps your backend up to date.

Hands-free Lovable sync: If you have the Staticbot Chrome extension installed, the agent can trigger Lovable deployments automatically via the local bridge. Without the extension, you'll paste deploy staticbot edge function into the Lovable AI chat — the only manual step in the whole pipeline.

What's New Since Launch

Self-hosted targets

Choose SUPABASE_SELF_HOSTED to get an AES-256-encrypted downloadable package instead of migrating directly to a managed project. Unzip, run Claude Code, follow the bundled CLAUDE.md.

Live preview deployments

Before switching over, Staticbot builds a preview of your app at a live URL — Cloudflare Workers for SSR apps, CloudFront for static ones — so you can verify everything works.

Managed hosting

After preview validation, promote to production hosting on Staticbot's infrastructure — Cloudflare Workers + Supabase or CloudFront + Supabase, matching your app type.

Self-navigating pipelines

The API returns a pendingAction telling the agent exactly what to do next — no hard-coded phase logic needed. Agents follow the breadcrumbs.

Why MCP?

Conversational

No need to learn a new UI. Your AI agent explains what's happening at each step and asks for approval when needed.

Faster setup

The agent reads your .env, finds your integrations, and creates the migration — no manual form-filling.

Smart error handling

If a job fails, the agent can retry or skip it. The API surfaces a categorized failureBanner with actionable explanations.

Works with any MCP client

Claude Code, Cursor, Windsurf, or any tool that supports the Model Context Protocol.

Ready to migrate from chat?

Get your API key, add the MCP server config, and tell your AI agent what to migrate. Your migration runs while you stay in your editor.

Get API Key MCP Server Docs