Use Supabase with TanStack Start

1. Create a Supabase project#

To start, you need a Supabase project.

Create a new Supabase project from the Dashboard of any organization you belong to.

2. Set up your database#

When your Supabase project is up and running, create an instruments table with some sample data. Then set only the privileges each Postgres role needs, add Row Level Security (RLS) for enhanced security for database data by default, and create an RLS policy to make the data in the table publicly readable.

Do these steps within your project's dashboard by copying and running the snippet in your project's SQL Editor.

1
-- Create the table
2
create table instruments (
3
  id bigint primary key generated always as identity,
4
  name text not null
5
);
6
7
-- Insert sample data into the table
8
insert into instruments (name)
9
values
10
  ('violin'),
11
  ('viola'),
12
  ('cello');
13
14
-- Grant the privileges the role needs, which is read access
15
grant select on public.instruments to anon;
16
17
-- Enable row level security for the table
18
alter table instruments enable row level security;
19
20
-- Create a policy to allow the anon role to read from the instruments table
21
create policy "public can read instruments"
22
on public.instruments
23
for select to anon
24
using (true);

3. Create a TanStack Start app#

Create a TanStack Start app using the official CLI.

1
npx @tanstack/cli@latest create my-app

4. Set up AI tooling (optional)#

Supabase provides two ways to give AI tools context about your project: Agent Skills, which give your AI coding agent procedural knowledge, and the MCP server, which connects AI assistants to your Supabase project directly.

Agent Skills#

Agent Skills is a curated set of instructions that give your AI agent procedural knowledge about working with Supabase.

Install them so your AI coding agent can produce more accurate, reliable code using current Supabase patterns, such as authentication, server-side rendering, and database migrations, rather than relying solely on training data.

Installing Agent Skills#

To install, run the following command in the root of your project:

1
npx skills add supabase/agent-skills

Supabase MCP server#

The Supabase MCP server connects AI assistants to Supabase, so they can inspect your schema and act on your projects on your behalf. Find out how to add it to your client in the MCP docs.

5. Install the Supabase client libraries#

Navigate to the TanStack Start app and install supabase-js and @supabase/ssr, the helper package that manages cookie-based sessions for server-side rendering.

1
cd my-app && npm install @supabase/supabase-js @supabase/ssr

6. Declare Supabase environment variables#

Create a .env.local file in the root of your project and populate it with your Supabase connection variables. Get the values from the helper below, or from the project Connect panel.

1
VITE_SUPABASE_URL=<SUBSTITUTE_SUPABASE_URL>
2
VITE_SUPABASE_PUBLISHABLE_KEY=<SUBSTITUTE_SUPABASE_PUBLISHABLE_KEY>

Get API details#

To interact with data in database tables, you use the client libraries that wrap the auto-generated Data API endpoints, authenticating using the Project URL and key from the project Connect dialog.

7. Create Supabase client utilities#

TanStack Start needs two Supabase clients: a browser client for components that run in the browser, and a server client for loaders and server functions. Create a src/lib/supabase folder with a file for each client.

Both clients read the same two variables, through the API available in each environment. The browser client uses import.meta.env, which Vite replaces at build time. The server client uses process.env, which the server runtime populates from your .env.local file.

1
/// <reference types="vite/client" />
2
import { createBrowserClient } from '@supabase/ssr'
3
4
export function createClient() {
5
  return createBrowserClient(
6
    import.meta.env.VITE_SUPABASE_URL!,
7
    import.meta.env.VITE_SUPABASE_PUBLISHABLE_KEY!
8
  )
9
}

1
import { createServerClient } from '@supabase/ssr'
2
import { getCookies, setCookie, setResponseHeader } from '@tanstack/react-start/server'
3
4
export function createClient() {
5
  return createServerClient(
6
    process.env.VITE_SUPABASE_URL!,
7
    process.env.VITE_SUPABASE_PUBLISHABLE_KEY!,
8
    {
9
      cookies: {
10
        getAll() {
11
          return Object.entries(getCookies()).map(([name, value]) => ({ name, value }))
12
        },
13
        setAll(cookies, headers) {
14
          cookies.forEach(({ name, value, options }) => {
15
            setCookie(name, value, options)
16
          })
17
18
          Object.entries(headers).forEach(([name, value]) => {
19
            setResponseHeader(name, value)
20
          })
21
        },
22
      },
23
    }
24
  )
25
}

8. Query Supabase data from TanStack Start#

Create a server function that queries the instruments table through the server client. TanStack Start's import protection blocks direct server imports in route files, so wrap the Supabase call in createServerFn.

1
import { createServerFn } from '@tanstack/react-start'
2
3
import { createClient } from '@/lib/supabase/server'
4
5
export const fetchInstruments = createServerFn({ method: 'GET' }).handler(async () => {
6
  const supabase = createClient()
7
  const { data: instruments, error } = await supabase.from('instruments').select()
8
9
  if (error) {
10
    console.error(error)
11
    return { instruments: [], error: error.message }
12
  }
13
14
  return { instruments, error: null }
15
})

Replace the contents of src/routes/index.tsx with the following to call the server function from a route loader. The loader runs on the server, so the data is part of the initial server-rendered response.

1
import { createFileRoute } from '@tanstack/react-router'
2
3
import { fetchInstruments } from '@/lib/supabase/fetch-instruments-server-fn'
4
5
export const Route = createFileRoute('/')({
6
  loader: async () => fetchInstruments(),
7
  component: Home,
8
})
9
10
function Home() {
11
  const { instruments, error } = Route.useLoaderData()
12
13
  if (error) {
14
    return <p>Error loading instruments: {error}</p>
15
  }
16
17
  return (
18
    <ul>
19
      {instruments?.map((instrument) => (
20
        <li key={instrument.id}>{instrument.name}</li>
21
      ))}
22
    </ul>
23
  )
24
}

9. Start the app#

Run the development server, go to http://localhost:3000 in a browser and you should see the list of instruments.

1
npm run dev

Production requirements#

The quickstart procedure in this guide optimizes for getting you to a working app, not for production.

Before you deploy:

• If your app reads or writes through the Data API, review your Row Level Security policies. Any policy you added here is scoped to this quickstart's sample data, not to real user data.
• Set your Supabase credentials as environment variables on whatever platform you deploy to, rather than committing them to source control.
• Configure a custom domain for your Supabase project once you're ready to go live.

Next steps#

• Learn how to protect routes and check sessions with the server client, or drop in a complete login and sign-up flow from Supabase Library
• Insert more data into your database
• Upload and serve static files using Storage
• Explore drop-in UI components for your Supabase app