Supabase

Work with Supabase projects through Supabase's official hosted MCP server. Discover the live tool catalog before calling a tool.

am-will Updated 1k repo stars
GooeyPi desktop workspace

File contents

Supabase MCP

Use Supabase's official hosted MCP server from the Prime Agent Python kernel.

Setup

Configure the supabase MCP server and complete /mcp login supabase directly in Prime Agent outside GooeyPi. GooeyPi intentionally does not forward MCP authentication commands or inspect shared MCP credentials. Prefer a project-scoped, read-only server URL for routine work.

Usage

Always discover the server-defined tools and schemas before calling them:

import supabase

tools = await supabase.list_tools()
print([(tool["name"], tool["description"]) for tool in tools])

Call a tool by its exact name and pass arguments matching its live input schema:

result = await supabase.call_tool("search_docs", {"graphql_query": "..."})

Every operation is async. Supabase MCP may expose database-changing tools unless the configured URL includes read_only=true; review each tool call and do not connect production data.

Official setup and security guidance: https://supabase.com/docs/guides/ai-tools/mcp

am-will/gooey-pi commit 2c3310a6d0

Frequently asked questions

npx get-skill add am-will/supabase