Connect Supabase to ChatGPT
Search, read and write Supabase from ChatGPT. Arlet links the two through one MCP connection and takes care of keys and authentication.
Arlet makes Supabase's official MCP server available from the AI client you already use. Signing in completes the connection.
The AI can manage projects, run SQL and migrations, read logs and advisories, deploy Edge Functions and search the docs. Write tools are included, so keep tool-call confirmation enabled in your AI client when working against production projects. The available tools are those provided by Supabase's official server.
How to connect Supabase to ChatGPT
- 1
In the Arlet dashboard, open Integrations β Add an integration and choose Supabase. Sign in with Link Supabase, name the connection and click Create.
- 2
In ChatGPT, open Settings β Plugins β Browse plugins and click + at the top right.
- 3
In the New Plugin form, enter "Arlet" as the Name, choose Server URL for Connection and enter https://mcp.arlet.dev/mcp/, pick OAuth for Authentication, check the risk acknowledgement and click Create.
- 4
Open the Arlet plugin you created, click Sign in with Arlet, sign in with Google and click Allow. The connection is complete.
If the Plugins screen is not shown, turn on Developer mode under Settings β Security and login.
What you can do
Supabase tools that ChatGPT can call.
list_projectsList the projects in your organization.
get_projectGet a project's details.
list_tablesList tables.
execute_sqlRun SQL.
apply_migrationApply a migration.
list_migrationsList applied migrations.
list_extensionsList enabled extensions.
query_logsQuery project logs.
get_advisorsGet security and performance advisories.
generate_typescript_typesGenerate TypeScript types from the schema.
list_edge_functionsList Edge Functions.
deploy_edge_functionDeploy an Edge Function.
get_project_urlGet the project's API URL.
get_publishable_keysGet the publishable API keys.
search_docsSearch Supabase's documentation.
The tool list follows what Supabase's official MCP server provides and may change.
Example uses
Investigate data
Inspect the schema, write and run aggregation queries, and explain the results.
Change the schema
Write a migration from requirements, apply it and regenerate TypeScript types.
Check logs and advisors
Search error logs and combine them with security and performance advisories into an action list.
Update an Edge Function
Fetch a function's code, fix it and deploy.
Prompts you can use as-is
- Count new sign-ups in the users table per day for the last 7 days
- Write and apply a migration adding a status text column with default 'pending' to orders
- Look through the last hour of API error logs and summarize likely causes
- List the security advisor findings and explain how to fix the highest-priority ones
