openlink/mcp-server-odbc

This document covers the set up and use of a generic ODBC server for the Model Context Protocol (MCP), referred to as an mcp-odbc server. It has been developed to provide Large Language Models with transparent access to ODBC-accessible data sources via a Data Source Name configured for a specific ODBC Connector (also called an ODBC Driver).
Server Implementation
This **MCP Server for ODBC** is a small TypeScript layer built on top of node-odbc. It routes calls to the host system's local ODBC Driver Manager via node.js (specifically using npx for TypeScript).
Operating Environment Set Up & Prerequisites
While the examples that follow are oriented toward the Virtuoso ODBC Connector, this guide will also work with other ODBC Connectors. We *strongly* encourage code contributions and submissions of usage demos related to other database management systems (DBMS) for incorporation into this project.
Key System Components
1. Check the node.js version. If it's not 21.1.0 or higher, upgrade or install explicitly using:
2. Install MCP components using:
3. Set the nvm version using:
Installation
1. Run
2. Change directory
3. Run
4. Run
unixODBC Runtime Environment Checks
1. Check installation configuration (i.e., location of key INI files) by running:
2. List available data source names (DSNs) by running:
Environment Variables
As good security practice, you should use the .env file situated in the same directory as the mcp-ser to set bindings for the ODBC Data Source Name (ODBC_DSN), the User (ODBC_USER), the Password (ODBC_PWD), the ODBC INI (ODBCINI), and, if you want to use the OpenLink AI Layer (OPAL) via ODBC, the target Large Language Model (LLM) API Key (API_KEY).
Usage
Tools
After successful installation, the following tools will be available to MCP client applications.
Overview
|name |description|
|:--- |:---|
|get_schemas |List database schemas accessible to connected database management system (DBMS).|
|get_tables |List tables associated with a selected database schema.|
|describe_table |Provide the description of a table associated with a designated database schema. This includes information about column names, data types, null handling, autoincrement, primary key, and foreign keys|
|filter_table_names |List tables associated with a selected database schema, based on a substring pattern from the q input field.|
|query_database |Execute a SQL query and return results in JSON Lines (JSONL) format.|
|execute_query |Execute a SQL query and return results in JSON Lines (JSONL) format.|
|execute_query_md |Execute a SQL query and return results in Markdown table format.|
|spasql_query |Execute a SPASQL query and return results.|
|sparql_query |Execute a SPARQL query and return results.|
|virtuoso_support_ai|Interact with the Virtuoso Support Assistant/Agent — a Virtuoso-specific feature for interacting with LLMs|
Detailed Description
- **get_schemas**
- Retrieve and return a list of all schema names from the connected database.
- Input parameters:
- user (string, optional): Database username. Defaults to "demo".
- password (string, optional): Database password. Defaults to "demo".
- dsn (string, optional): ODBC data source name. Defaults to "Local Virtuoso".
- Returns a JSON string array of schema names.
- **get_tables**
- Retrieve and return a list containing information about tables in a specified schema. If no schema is provided, uses the connection's default schema.
- Input parameters:
- schema (string, optional): Database schema to filter tables. Defaults to connection default.
- user (string, optional): Database username. Defaults to "demo".
- password (string, optional): Database password. Defaults to "demo".
- dsn (string, optional): ODBC data source name. Defaults to "Local Virtuoso".
- Returns a JSON string containing table information (e.g., TABLE_CAT, TABLE_SCHEM, TABLE_NAME, TABLE_TYPE).