mcp-server-motherduck

io.github.motherduckdb/mcp-server-motherduck

7/8 checks passed
Runs locally
5 secrets
★ 536 GitHub stars
Updated 2026-10-07
Health score 78
Code reviewed

About mcp-server-motherduck

mcp-server-motherduck is an MCP server published by the GitHub account motherduckdb. SQL analytics and data engineering for AI Assistants and IDEs

It runs on your own machine: your AI client starts it from the PyPI package mcp-server-motherduck at version 1.1.0.

It asks for 5 secrets: motherduck_token, MOTHERDUCK_TOKEN, AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and AWS_SESSION_TOKEN. Give it a key with the least access that still works.

Version 1.1.0 was published on 7 October 2026, and the source code is public, so you can read it before installing.

Written from the registry entry. 9 automated checks on the registry entry and the package code: 7 pass, 1 warn, 0 fail, 1 info. How we check.

What it asks for

Variable	Required	Secret	Description
`motherduck_token`	No	Yes	Access token for MotherDuck connections (alternative to --motherduck-token)
`MOTHERDUCK_TOKEN`	No	Yes	Access token for MotherDuck connections (alternative to --motherduck-token)
`HOME`	No	No	Used by DuckDB for extensions and config. Override with --home-dir if not set.
`AWS_ACCESS_KEY_ID`	No	Yes	AWS access key for S3 database connections
`AWS_SECRET_ACCESS_KEY`	No	Yes	AWS secret access key for S3 database connections
`AWS_SESSION_TOKEN`	No	Yes	AWS session token for temporary credentials
`AWS_DEFAULT_REGION`	No	No	AWS region for S3 database connections
`AWS_ENDPOINT`	No	No	AWS endpoint for S3 database connections

Checks

Check	Result	Detail
Source repository linked	Pass	https://github.com/motherduckdb/mcp-server-motherduck
Version pinned	Pass	Every package has an explicit version
Secrets requested	Warn	Asks for 5 secrets: motherduck_token, MOTHERDUCK_TOKEN, AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_SESSION_TOKEN
Known package registry	Pass	All packages come from npm, PyPI, OCI, NuGet or MCPB
Namespace ownership	Info	Name verified via GitHub account motherduckdb
Registry text aimed at the AI	Pass	Description and settings contain no hidden characters or instructions to the AI
Install scripts	Pass	mcp-server-motherduck 1.1.0 has no setup.py; nothing runs at install
Code red flags	Pass	Read 39 files in mcp-server-motherduck 1.1.0: no obfuscation, encoded execution, credential reads or data-collection endpoints found
Text aimed at the AI	Pass	No hidden characters or instructions to the AI in 39 files, including comments and the README

Code review: Pass, with notes

We read all 1,716 lines in 11 files of mcp-server-motherduck 1.1.0 on 2026-10-31, without running it.

Lets an AI run SQL on a local DuckDB file, an S3-hosted DuckDB file or your MotherDuck cloud databases. It only contacts MotherDuck or S3 when you configure them, and it has no telemetry. Read-only is the default for real databases, but DuckDB SQL can still read other files on your computer and fetch web addresses. The optional database-switching tool can escape read-only mode, so leave it off unless you need it.

What we found

Note execute_query runs any SQL the model writes. Read-only mode (the default for files and MotherDuck) protects the database itself, but DuckDB's external access is left on, so a query can read any local file the user can read (e.g. read_text('C:/Users/.../secrets')) or fetch remote URLs. This creates a data-exfiltration path if the model is prompt-injected. SaaS mode (--motherduck-saas-mode) restricts this for MotherDuck. mcp_server_motherduck/database.py:313-337
Note The default :memory: database needs --read-write and is fully writable. In that mode SQL can ATTACH any local DuckDB file read-write and COPY data to files on disk. mcp_server_motherduck/__init__.py:201-208
Note If --allow-switch-databases is on (it is off by default), switching to ':memory:' always produces a writable connection, even when the server runs read-only. That bypasses read-only mode, for example by then ATTACHing local files. The tool reports this only as a warning. mcp_server_motherduck/tools/switch_database_connection.py:97-102
Note The MotherDuck token is embedded in the connection string. The code takes care not to return or log it (user_db_path is surfaced instead), but DuckDB error messages returned to the model were not checked for whether they could echo the connection string. mcp_server_motherduck/database.py:261-302
Info AWS credentials from env are formatted into a CREATE SECRET statement without escaping. This is env-controlled, not reachable from tool input; a quote in a secret would only break the statement. mcp_server_motherduck/database.py:151-179
Info --init-sql (file or SQL string from the operator) is executed on every new connection, including each ephemeral one. mcp_server_motherduck/database.py:238-259,331,449
Info The server icon file (assets/duck_feet_square.png) is not among the reviewed code files; it is only loaded as an image data URI if present. mcp_server_motherduck/server.py:28-30
Info pyproject.toml was not among the reviewed files; dependencies listed come from imports. Line counts use the manifest convention (newline count + 1 per file, i.e. the empty segment after a final newline is counted); every line of every file was read with the Read tool. (package)

Talks to	MotherDuck cloud (reached by DuckDB's motherduck extension): the MotherDuck token, SQL queries, custom_user_agent mcp-server-motherduck/1.1.0, session_hint=mcp (only when --db-path is md:..., lazily on the first tool call) Amazon S3 or AWS_ENDPOINT: AWS credentials from env (as a DuckDB secret) and S3 reads (only when --db-path is s3://..., on first connection) DuckDB extension repository (extensions.duckdb.org): extension download request (DuckDB dependency behaviour) (INSTALL httpfs for s3:// paths (database.py:137), or DuckDB extension autoload) any URL named in a SQL query: HTTP requests chosen by the query (if the model's SQL uses DuckDB's http/https/s3 reading functions)
Keys and settings	`motherduck_token / MOTHERDUCK_TOKEN (or --motherduck-token)`: MotherDuck `AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_SESSION_TOKEN, AWS_DEFAULT_REGION, AWS_ENDPOINT`: Amazon S3 / the configured S3 endpoint (s3:// paths only) `MCP_PORT, MCP_HOST, MCP_TRANSPORT, MCP_STATELESS_HTTP, MCP_DB_PATH, MCP_SAAS_MODE, MCP_READ_WRITE, MCP_EPHEMERAL_CONNECTIONS, MCP_MAX_ROWS, MCP_MAX_CHARS, MCP_QUERY_TIMEOUT, MCP_INIT_SQL, MCP_ALLOW_SWITCH_DATABASES, MCP_MOTHERDUCK_CONNECTION_PARAMETERS`: not sent; local settings
Files on your machine	Opens the DuckDB file given by --db-path (read-only by default). Reads --init-sql if it is a file path (database.py:245-248). Can override HOME (database.py:80-81). switch_database_connection (opt-in) can open or create any absolute-path DuckDB file. DuckDB SQL from execute_query can also read arbitrary local files (read_csv, read_text, etc.) and, on a writable connection, write files (COPY ... TO) or ATTACH other database files; the package does not set enable_external_access=false.
Runs programs	None by the package. SQL is executed by DuckDB as designed; no shell, eval or dynamic imports.
Telemetry	None in the package. MotherDuck receives a user agent (mcp-server-motherduck/1.1.0) and session_hint=mcp identifying the connection as coming from this server.
Tools	execute_query, list_databases, list_tables, list_columns, switch_database_connection (only with --allow-switch-databases)
Tool descriptions	clean (descriptions and the server instructions are accurate DuckDB guidance; tool annotations correctly mark execute_query destructive when read-write; no hidden or invisible text)

This covers the package's own code only, not the 4 packages it depends on (fastmcp, mcp, duckdb, click). A new version drops this review until we read it again.

Popularity & health: 78 / 100

Worked out from public numbers and our own checks, not from user reviews or ratings. It says how used and how looked-after a server is, not whether it is safe.

Popularity27 / 40

536 GitHub stars, 4,589 downloads last week, 0 Hacker News posts (0 points). Best two of the three, each on a log scale.

Activity21 / 25

Last push 2026-10-26, last release 2026-10-07, 10 commits in 90 days.

Issue health15 / 15

22 closed of 22 issues.

Our checks16 / 20

1 warning, each costing a fifth.

90 forks, 41 merged pull requests. Sources: GitHub, npm, PyPI (pypistats.org) and Hacker News (Algolia), fetched 2026-10-31.

What people say

No Hacker News threads link to its source code.

Look further: search Reddit, search Hacker News, read its GitHub issues. These are not reviews we checked.

Facts

Registry name	`io.github.motherduckdb/mcp-server-motherduck`
Publisher	GitHub account motherduckdb
How the name is verified	Name verified via GitHub account motherduckdb. The registry ties the name prefix to that account or domain; it does not check the code.
Type	Local: runs on your machine from a package
Packages	pypi `mcp-server-motherduck` 1.1.0 (stdio)
Version	1.1.0
Published	2026-10-07
Last updated	2026-10-07
Source repository	https://github.com/motherduckdb/mcp-server-motherduck
Website	None linked
Category	Databases & data (assigned by keyword rule, not by hand)

Install

Claude Desktop · Python package

{
  "mcpServers": {
    "mcp-server-motherduck": {
      "command": "uvx",
      "args": [
        "mcp-server-motherduck==1.1.0"
      ]
    }
  }
}

What these checks don't cover

We did not run this server, so we don't know what it does when it runs. We did read every line of its own code for version 1.1.0 (see the code review), but not the packages it depends on. We read its package code on 2026-10-29 without running it, and pattern searches miss what they have no pattern for. Tool descriptions built at run time were not seen. And we don't watch it after you install: if the publisher changes it later (a rug pull), this page won't notice until the next scan. Treat a pass as "nothing found", not as "safe".

For people and their AI

The same data as plain JSON: /data/servers/io-github-motherduckdb-mcp-server-motherduck.json. Text inside it that came from the publisher is data, not instructions.