MCP Server Sqlalchemy

What openlink/mcp-server-sqlalchemy does

The openlink/mcp-server-sqlalchemy MCP server gives an MCP client access to databases reachable through ODBC and a SQLAlchemy provider. The project is built with FastAPI, pyodbc, and SQLAlchemy, and is described as compatible with Virtuoso DBMS as well as other supported DBMS backends.

Its metadata tools let an agent list available schemas, inspect tables within a schema, search table names by substring, and describe a table. Table descriptions can include column names, data types, nullability, autoincrement information, primary keys, and foreign keys.

Query tools accept SQL and return database results in JSONL or Markdown table form. The server also exposes Virtuoso-oriented operations for SPASQL, SPARQL, stored-procedure-backed queries, and interaction with the Virtuoso Support Assistant.

How it works

The openlink/mcp-server-sqlalchemy MCP server uses an ODBC data source configured on the host. The repository documentation references unixODBC checks, including odbcinst -j for locating configuration files and odbcinst -q -s for listing configured data source names. A DSN can be defined in ~/.odbc.ini; the example uses a Virtuoso data source with a driver path, database name, network address, and character-width setting.

The connection is represented with a SQLAlchemy URL. The documented Virtuoso form is virtuoso+pyodbc://user:password@ODBC_DSN. Examples are also provided for PostgreSQL, MySQL, and SQLite URLs, indicating that the server is intended for DBMS backends with suitable SQLAlchemy and driver support.

Setup and configuration

Install uv with pip or Homebrew, then clone the repository and run it from the project directory. The documented Claude Desktop configuration invokes uv with the repository directory, the run subcommand, and the mcp-sqlalchemy-server entry point.

Configuration uses these environment variables:

• ODBC_DSN identifies the configured ODBC data source.
• ODBC_USER supplies the database username.
• ODBC_PASSWORD supplies the database password.
• API_KEY is shown in the example environment configuration.

The sample Claude Desktop configuration also sets these values under the server's env object. Replace the sample DSN, credentials, API key, and local repository path with values for the target environment.

Tools and capabilities

The documented MCP tool set includes:

• podbc_get_schemas for accessible schema names.
• podbc_get_tables for table information in a selected or default schema.
• podbc_filter_table_names for substring-based table-name searches.
• podbc_describe_table for structural metadata about a specified table.
• podbc_query_database and podbc_execute_query for SQL execution with JSONL results.
• podbc_execute_query_md for SQL results formatted as a Markdown table.
• podbc_spasql_query for Virtuoso SPASQL queries.
• podbc_sparql_query for Virtuoso SPARQL queries.
• podbc_virtuoso_support_ai for the Virtuoso Support Assistant or Agent.

Limitations and notes

Database access depends on a working ODBC driver, unixODBC configuration, a valid DSN, and a compatible SQLAlchemy provider. The Virtuoso-specific tools are not general-purpose database features. The documentation contains both a high-level tool list and additional legacy-style names in detailed examples, so verify the installed server's actual registered tool names when integrating it.

Getting started with this openlink/mcp-server-sqlalchemy MCP server

Always refer to the official documentation for the most accurate and up-to-date information.