get_allowed_databasesLABSCO SUMMARY
Reads and writes are different tools, and the database list is set before start-up.
--allowed-databases fixes what the server can reach when it launches, so touching the wrong database takes a config change rather than a bad guess. Splitting read_query from write_query means a client that approves tools individually can allow the first and hold the second.
What it is
A Python server for Microsoft SQL Server Express. Reads and writes are separate tools, and the databases it may touch are named on the launch command rather than chosen per call.
What you get
- read_query for SELECT
- write_query for INSERT, UPDATE and DELETE
- create_table to add a table
- list_tables for what exists, describe_table for a table's schema
- get_allowed_databases for the databases this instance was launched with
- Windows authentication or SQL Server authentication, chosen at launch
Requirements
Python 3.10 or newer, Microsoft's ODBC driver for SQL Server, and an instance you have rights on. Launch arguments carry the configuration: --server, --auth, and --allowed-databases as a comma-separated list. SQL authentication adds a user name and a password on the same line; Windows authentication uses --trusted-connection yes.
Setup effort
Build from source — get the source, then follow the steps on this page and add the configuration for your client
USE CASES
01
Answering a question against a database nobody documented
List the tables, read the schema of the ones that look right, and then write the SELECT.
02
Letting an assistant read production without letting it write
Reads and writes are separate tools, so approve read_query and leave write_query unapproved in the client.
03
Applying a data fix you have already agreed on
Run the UPDATE or DELETE directly instead of pasting it into a client.
04
Confirming which databases this connection can even see
Ask for the allowed list — it is fixed at launch, so the answer is definitive.
05
Standing up a table for something you are prototyping
Create it in one of the allowed databases without leaving the conversation.
6 tools
Listed in project docs. Your client may show a different list.
read_querywrite_querycreate_tablelist_tablesdescribe_tableBEFORE YOU INSTALL
What to know first
Tools you get
6 tools named in the project's documentation.
Requires
- uv on PATH
- Python 3.10 or later
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Get the source, then follow the steps on this page and add the configuration for your client. This one is built from source rather than installed with a single command.
Server setup
From source
TERMINAL
git clone https://github.com/hanweg/mcp-sqlexpress
cd mcp-sqlexpress
# Create and activate virtual environment
uv venv
.venv\Scripts\activate
# Install dependencies
uv pip install --editable .uv
TERMINAL
uv --directory <YOUR_PATH_TO_PROJECT_MCP_SQLEXPRESS> run mcp-server-sqlexpress --server server\instance --auth windows --trusted-connection yes --trust-server-certificate yes --allowed-databases database1,database2MCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"sqlexpress": {
"command": "uv",
"args": [
"--directory",
"<YOUR_PATH_TO_PROJECT_MCP_SQLEXPRESS>",
"run",
"mcp-server-sqlexpress",
"--server",
"server\\instance",
"--auth",
"windows",
"--trusted-connection",
"yes",
"--trust-server-certificate",
"yes",
"--allowed-databases",
"database1,database2"
]
}
}
}
