DuckDB MCP Server
A Model Context Protocol (MCP) server that gives AI assistants full access to — query local files, S3 buckets, and in-memory data using plain SQL.
Potencialize seu agente
Instalação do publicador. Não é um endpoint do PlugMCP.
Host a copy with our keys and logs.
Ainda não há uma versão empacotada no registro. Siga o README do repositório.
URL remota ou um comando na sua máquina? Como as duas instalações diferem
Rodar no PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
AbrirDo publicador
A Model Context Protocol (MCP) server that gives AI assistants full access to — query local files, S3 buckets, and in-memory data using plain SQL.
MCP (Model Context Protocol) server for DuckDB — auth and friendly SQL support out of the box
This server exposes DuckDB to any MCP-compatible client (Claude Desktop, Cursor, VS Code, etc.). The assistant can:
GROUP BY ALL, SELECT * EXCLUDE, etc.)Three built-in XML documentation resources are always available to the assistant:
pip install duckdb-mcp-server
From source:
git clone https://github.com/mustafahasankhan/duckdb-mcp-server.git
cd duckdb-mcp-server
pip install -e .
duckdb-mcp-server --db-path <path> [options]
Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"duckdb": {
"command": "duckdb-mcp-server",
"args": ["--db-path", "~/claude-duckdb/data.db"]
}
}
}
{
"mcpServers": {
"duckdb": {
"command": "duckdb-mcp-server",
"args": [
"--db-path", "~/claude-duckdb/data.db",
"--s3-region", "us-east-1",
"--creds-from-env"
],
"env": {
"AWS_ACCESS_KEY_ID": "YOUR_KEY",
"AWS_SECRET_ACCESS_KEY": "YOUR_SECRET"
}
}
}
}
Useful when the database is shared or managed separately:
{
"mcpServers": {
"duckdb": {
"command": "duckdb-mcp-server",
"args": ["--db-path", "/shared/analytics.db", "--readonly"]
}
}
}
"Load sales.csv and show me the top 5 products by revenue."
SELECT
product_name,
SUM(quantity * price) AS revenue
FROM read_csv('sales.csv')
GROUP BY ALL
ORDER BY revenue DESC
LIMIT 5;
"Cache this month's signups from S3 and show me a daily breakdown."
-- Step 1: cache the remote data locally
CREATE TABLE signups AS
SELECT * FROM read_parquet('s3://my-bucket/signups/2026-03/*.parquet',
union_by_name = true);
-- Step 2: query the cached table
SELECT
date_trunc('day', signup_at) AS day,
COUNT(*) AS signups
FROM signups
GROUP BY ALL
ORDER BY day DESC;
"Give me a statistical summary of the orders table."
The assistant calls analyze_data("orders") which returns row count, numeric stats per column, date ranges, and top categorical values — no SQL required from you.
--creds-from-env): AWS_ACCESS_KEY_ID + AWS_SECRET_ACCESS_KEY--s3-profile): reads ~/.aws/credentialsgit clone https://github.com/mustafahasankhan/duckdb-mcp-server.git
cd duckdb-mcp-server
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
pytest
Contributions are welcome! Please feel free to submit a Pull Request. MIT
MCP (Model Context Protocol) server for DuckDB — auth and friendly SQL support out of the box
Ferramentas
Ainda não há lista de ferramentas. Servidores remotos recebem uma após a primeira verificação de saúde.
Perguntas frequentes
Copie nesta página a configuração stdio para Cursor, Claude Desktop, Claude Code ou VS Code. Ela usa a instalação do publicador, não um proxy do PlugMCP.
Este registro não lista nenhuma variável de ambiente obrigatória.
Duckdb roda localmente via stdio.
Última verificação: 2026-10-28. MCP (Model Context Protocol) server for DuckDB — auth and friendly SQL support out of the box