Claude Code Supabase Integration

Authenticate the Supabase CLI

Required before Claude Code can create migrations, push schema changes, or generate types.

Link the repo to a project

Connects local migrations and config to the correct remote Supabase project.

Create a new migration

Creates a timestamped SQL file under supabase/migrations for Claude Code to fill in and you to review.

Apply migrations to the database

Runs pending migrations against the linked project. Review the SQL and test on staging before pushing to production.

Generate TypeScript types from the schema

Regenerate after any migration so the compiler catches a schema mismatch before it ships.

Deploy an edge function

Deploys a Deno edge function. Store secrets as environment variables, never in the function code.

I am building the following feature. Design a normalized Postgres schema for it in Supabase: the tables, columns and types, primary and foreign keys, useful indexes, and any constraints. Explain the trade-offs and show the SQL, but do not run anything yet.

Read my current Supabase schema through the MCP server and review it. Point out missing indexes, denormalization or normalization issues, columns that should be non-nullable or constrained, and any tables that do not have Row Level Security enabled.

Based on the current schema, write a Supabase migration that makes the following change. Produce the SQL for a new file under supabase/migrations, keep it idempotent where possible, and explain what it does. I will review before running supabase db push.

Enable Row Level Security on this table and write the policies so that: each user can only read and modify their own rows, and inserts must set the user id to the authenticated user. Show the SQL, then explain in plain language exactly what each policy allows and denies.

Review the Row Level Security policies across my Supabase tables. Identify any table with RLS disabled, any policy that could expose another user's rows, and any table where insert or update is unprotected. For each finding, explain the risk and the fix.

Write a SQL query for my Supabase database that returns the following result. Use the real table and column names from my schema, include the necessary joins and filters, and explain how the query works.

This query is slow. Here is the query and its EXPLAIN ANALYZE output. Identify why it is slow, tell me which index would help, and rewrite the query if a better formulation exists. Explain the reasoning before suggesting any change.

Help me set up Supabase Auth in this project for email and OAuth sign-in. Show the client setup, correct session handling, how to protect routes, and how to pair the auth rules with matching RLS policies so access is enforced at the database.

I need a storage bucket for user avatars where each user can upload and replace their own avatar and anyone can read them. Create the bucket configuration and the storage policies, and explain who can upload, read, and delete.

Write a Supabase edge function that handles an incoming Stripe webhook: verify the signature, handle the relevant event types, and update the database accordingly. Read secrets from environment variables, never hardcode them, and show me how to deploy it.

Show me the exact supabase CLI command to generate TypeScript types from my linked project, and how to wire the generated types into the Supabase client so my queries are type-safe. Then remind me to regenerate after each migration.

Generate a seed script for my Supabase database that inserts realistic sample data across these tables while respecting all foreign keys and constraints. The data should be varied enough to exercise the app locally and in CI.

Set up a realtime subscription on this table so my client reacts to inserts, updates, and deletes. Show the Supabase client code, explain how to filter events, and note any RLS or replication settings the table needs for realtime to work.

My app runs in a serverless environment. Explain whether I should use the Supabase connection pooler or a direct connection, why, and give me the correctly formatted connection string for the right one. Note any settings that matter for serverless.

I want to split this table into two related tables without downtime. Plan the change as an ordered series of migrations that keep the app working at each step, including how to backfill and cut over, and flag the highest-risk migration.

Here is a Row Level Security policy. Explain in plain language exactly which rows it allows each role to read or write, walk through an example for an authenticated user and an anonymous user, and tell me if it does anything I might not expect.

supabase db push failed with the following error. Explain what the migration was trying to do, why it failed, and the smallest correct fix. If the failure could have left the database in a partial state, tell me how to check and recover.

Given these frequent queries against my Supabase tables, recommend the indexes that would speed them up. Explain the trade-off in write cost and storage for each index, and write the migration to add the ones that are clearly worth it.

Read my Supabase schema and generate documentation for it: a short description of each table, its columns and relationships, and a summary of its Row Level Security policies. Write it so a new contributor can understand the backend without reading the SQL.

Review this Supabase edge function before I deploy it. Check for hardcoded secrets, missing input validation, unhandled errors, and any database call that bypasses RLS with the service_role key. For each issue, explain the risk and the fix.

What is the Claude Code Supabase integration?

It is a workflow that lets Claude Code work directly with your Supabase backend. Claude Code can design your Postgres schema, write and run migrations, generate Row Level Security policies, write queries, configure auth and storage, build edge functions, and generate TypeScript types, using your real database as the source of truth. It enhances your existing Supabase workflow rather than replacing it.

How do I connect Claude Code to Supabase?

There are two paths. Connect the official Supabase MCP server so Claude Code can inspect tables, run read-only queries, and read project configuration in context. Or install and authenticate the supabase CLI with supabase login and link your project, so Claude Code can create migrations, run supabase db push, and generate types. Many developers use both together.

Can Claude Code write SQL migrations for Supabase?

Yes. Claude Code reads your current schema, then writes a migration file under supabase/migrations that adds the tables, columns, or constraints you asked for. You review the generated SQL, test it against a local or staging database, and apply it with supabase db push when you are satisfied it is correct.

Can Claude Code generate Row Level Security policies?

Yes. Describe an access rule in plain English, such as users can only read their own rows, and Claude Code enables RLS and writes the select, insert, update, and delete policies. It can also explain exactly which rows each policy exposes and audit your existing policies for gaps before you rely on them.

Is it safe to give Claude Code database access?

It is safe when you follow least privilege. Scope the MCP access token to a single project, prefer read-only access for inspection, and keep the service_role key server-side only. Claude Code produces reviewable SQL that you approve before it runs, so a human stays in control of anything that touches production data.

Can Claude Code generate TypeScript types from my schema?

Yes. Claude Code can run supabase gen types typescript against your linked project to produce types that match the live schema, then wire them into the Supabase client so your queries are type-safe. Regenerate the types after each migration so the compiler catches any mismatch before it ships.

Can Claude Code build Supabase edge functions?

Yes. Claude Code can scaffold and write a Deno-based edge function for a webhook, scheduled job, or server-side logic, and deploy it with supabase functions deploy. It keeps secrets in environment variables rather than in the function code, which is essential because edge functions often use privileged keys.

Should I use the connection pooler or a direct connection?

Use the connection pooler for serverless or high-concurrency environments where many short-lived functions each open a connection, since the pooler prevents connection exhaustion. Use a direct connection for long-lived servers and for migration tooling. Claude Code can tell you which to use for your setup and give you the correctly formatted connection string.