mcp-jdbc

This MCP server implements a JDBC-based data access layer for relational databases (notably Virtuoso and other JDBC-backed DBMS). It exposes a range of MCP client tools that let you list schemas and tables, describe table structures, filter tables by name, and run SQL or Virtuoso-specific queries. Key capabilities include jdbc_get_schemas, jdbc_get_tables, jdbc_describe_table, jdbc_filter_table_names, jdbc_query_database, jdbc_query_database_md, jdbc_query_database_jsonl, and Virtuoso-specific features like jdbc_spasql_query and jdbc_sparql_query. Use these tools from your MCP client to discover database structure, execute queries, and retrieve results in JSONL or Markdown formats, depending on the tool you invoke.

Prerequisites:

• Java 21 or above
• A JVM environment suitable for running the MCPServer-1.0.0-runner.jar
• A JDBC-compatible database (Virtuoso or any other DBMS with a JDBC driver) and credentials

Installation steps:

1. Clone the repository: git clone https://github.com/OpenLinkSoftware/mcp-jdbc-server.git cd mcp-jdbc-server
2. Ensure you have Java 21+ installed and available in your PATH.
3. Obtain or build the MCP server jar (the project provides MCPServer-1.0.0-runner.jar as the runnable artifact).
4. Create or update the environment configuration to point to your database:
   • Set jdbc.url, jdbc.user, jdbc.password, and any required jdbc.api_key in an environment file or container environment variables.
5. Run the MCP server using the Java command from the mcp_config that will point to MCPServer-1.0.0-runner.jar. For example: java -jar MCPServer-1.0.0-runner.jar
6. Verify the MCP client can connect to the running server and list available tools (jdbc_get_schemas, jdbc_get_tables, etc.).

Environment variables:

• jdbc.url: JDBC connection string (e.g., jdbc:virtuoso://localhost:1111)
• jdbc.user: Database username
• jdbc.password: Database password
• jdbc.api_key: Optional API key for authentication if required by your setup

Tips:

• Ensure the JDBC driver for your target DBMS is accessible to the MCP server classpath if you’re using a custom driver setup.
• For Claude Desktop users, you can embed the Java command with -jar path to MCPServer-1.0.0-runner.jar and provide jdbc.* env values in claude_desktop_config.json as shown in the README.
• When using jdbc_query_database_jsonl or other JSONL outputs, you can stream results to clients that process line-delimited JSON.
• The Virtuoso-specific features (jdbc_spasql_query, jdbc_sparql_query) require a Virtuoso backend and appropriate query strings to return results.