Tibero

This MCP server enables secure communication between Tibero databases and AI applications by exposing database resources, executing SQL, and providing schema and sample data through a consistent Model Context Protocol interface. It focuses on safe read/write operations and schema insights, making it easier to build AI-driven data applications on Tibero.

How to use

You run the Tibero MCP Server locally and connect your MCP client to it to access Tibero resources. You can browse tables and views as MCP resources, run SQL queries (with proper safeguards), and retrieve schema details and sample data. Use the provided tools to execute queries, inspect table information, and fetch metadata about your Tibero structures. When you issue data-modifying statements, ensure you explicitly commit and manage transactions as controlled by your client.

How to install

Prerequisites you need before starting:

• Python 3.11 or higher
• Java Runtime Environment 8 or higher
• Tibero JDBC driver tibero6-jdbc.jar (included in the drivers path)
• A Tibero database accessible from your environment Follow these steps to install and prepare the MCP server environment.

Additional sections

Configuration and runtime details are provided below to help you get started quickly and securely.

To run the server locally, you will start a Python process that hosts the MCP server script. The runtime keeps AutoCommit off for DML to ensure explicit control over transactions.

pip install -r requirements.txt

python src/tibero_mcp_server/server.py

python -m tibero_mcp_server.server

Configuration example

{
  "mcpServers": {
    "tibero": {
      "command": "python",
      "args": ["/path/to/tibero_mcp_server/src/tibero_mcp_server/server.py"],
      "env": {
        "TIBERO_HOST": "localhost",
        "TIBERO_PORT": "8629",
        "TIBERO_SID": "tibero",
        "TIBERO_USER": "username",
        "TIBERO_PASSWORD": "password",
        "CLASSPATH": "/path/to/tibero6-jdbc.jar"
      }
    }
  }
}

Security and notes

Only select operations are recommended for typical usage beyond reading data due to safety considerations. Always validate inputs and manage access control in your MCP client to prevent unauthorized data modifications.

Examples of available actions

You can use the following tools to interact with Tibero resources through MCP.

Tools

execute_sql runs SQL queries and returns results in CSV format for SELECT/SHOW/DESC, or the number of affected rows with an explicit commit for INSERT/UPDATE/DELETE, and returns outcomes for DDL statements.

Data access resources

Resources expose Tibero data through these paths:

• Tables: tibero://TABLE_NAME/data – table schema plus up to 100 sample rows
• Views: tibero://VIEW_NAME/view – view schema plus sample data

Table information

get_table_info retrieves detailed table metadata, including column names, types, lengths, NULL permissions, constraints (PRIMARY KEY, UNIQUE, FOREIGN KEY, CHECK), and index information (name, uniqueness). It helps you understand the exact structure you are querying.

Available tools

execute_sql

Executes SQL queries. Returns CSV for reads and the count of affected rows for write operations, with explicit commit for DML and a results payload for DDL.

get_table_info

Fetches detailed metadata about a table, including columns, constraints, and indexes.