You can run an MCP server that exposes ODBC access through PyODBC, enabling you to query and describe database schemas, tables, and stored procedures from your MCP client. This server is designed to work with Virtuoso and other ODBC-backed databases, providing JSONL, Markdown, and other formats for query results.

How to use

You connect to the ODBC MCP server from your MCP client by configuring the client to talk to a locally running server or a remote MCP endpoint. Once connected, you can fetch available schemas, list tables within schemas, describe table structures, filter tables by name, and run standard SQL or SPASQL-like queries when supported. Use the client’s commands to invoke the endpoints exposed by the server, and choose the desired result format such as JSONL or Markdown for easy downstream consumption.

How to install

pip install uv

brew install uv

odbcinst -j

odbcinst -q -s

# Configure your ODBC DSN in ~/.odbc.ini for Virtuoso or your target DB
[VOS]
Description = OpenLink Virtuoso
Driver = /path/to/virtodbcu_r.so
Database = Demo
Address = localhost:1111
WideAsUTF16 = Yes

# Then clone the MCP server repository and prepare to run

git clone https://github.com/OpenLinkSoftware/mcp-pyodbc-server.git
cd mcp-pyodbc-server

# Create or edit environment settings for the DSN and credentials

ODBC_DSN=VOS ODBC_USER=dba ODBC_PASSWORD=dba API_KEY=xxx

Optional: start an inspector to troubleshoot MCP interactions (if needed)

npm install -g @modelcontextprotocol/inspector
npx @modelcontextprotocol/inspector uv --directory /path/to/mcp-pyodbc-server run mcp-pyodbc-server

Configuration

{
  "mcpServers": {
    "pyodbc_mcp": {
      "command": "uv",
      "args": ["--directory", "/path/to/mcp-pyodbc-server", "run", "mcp-pyodbc-server"],
      "env": {
        "ODBC_DSN": "dsn_name",
        "ODBC_USER": "username",
        "ODBC_PASSWORD": "password",
        "API_KEY": "sk-xxx"
      }
    }
  }
}

Additional configuration notes

The server provides a set of tools to interact with the connected database. You can fetch schemas, list and describe tables, filter table names by a substring, and execute queries in multiple formats. If you need Virtuoso-specific features, you can also run SPASQL queries or use the Virtuoso AI assistant functionality where supported.

Troubleshooting

# Install MCP Inspector for quick troubleshooting
npm install -g @modelcontextprotocol/inspector

npx @modelcontextprotocol/inspector uv --directory /path/to/mcp-pyodbc-server run mcp-pyodbc-server

# Visit the troubleshooting URL provided by the inspector to diagnose issues

Available tools

podbc_get_schemas

Retrieve and return a list of all schema names from the connected database.

podbc_get_tables

Retrieve and return information about tables in a specified schema or the default schema.

podbc_filter_table_names

Filter and return tables whose names contain a given substring.

podbc_describe_table

Provide a detailed description of a table's columns, including data types and constraints.

podbc_query_database

Execute a SQL query and return the results in JSON format.

podbc_query_database_md

Execute a SQL query and return the results formatted as a Markdown table.

podbc_query_database_jsonl

Execute a SQL query and return the results in JSON Lines format.

podbc_spasql_query

Execute a SPASQL query (Virtuoso-specific) and return results.

podbc_virtuoso_support_ai

Interact with Virtuoso AI assistant for support queries using a prompt.