MySQL MCP Server

​

Features

​

Database Connection Management

• Load connection configs from JSON or environment variables
• List all configured database connections
• Validate connectivity and resolve provider mode (MySQL AI vs. MySQL HeatWave)

​

Database Operations

• Execute SQL queries with parameterized statements
• Support for multiple connection configurations

​

MySQL AI and HeatWave ML/GenAI

• Text Generation: Generate text using GenAI models
• Vector Embeddings: Create and populate vector columns for semantic search
• RAG (Retrieval-Augmented Generation): Query vector stores with natural language
• NL2SQL: Convert natural language questions into SQL queries and execute automatically
• HeatWave Help: Get answers about using HeatWave ML features

​

Vector Store Management

• List files in secure_file_priv directory (local mode)
• Load documents into vector stores from local filesystem
• Load documents from OCI Object Storage into vector stores

​

OCI Object Storage Tools

• List all compartments in your tenancy
• Get compartment by name
• List buckets in a compartment
• List objects in a bucket

​

MySQL AI vs MySQL HeatWave

​

MySQL AI

• Cloud-native MySQL service with integrated AI capabilities
• Runs on MySQL Database Service instances
• Vector processing happens within the database service
• Supports local filesystem for document loading
• Works without OCI config for pure database operations

​

MySQL HeatWave

• In-memory analytics engine on Oracle Cloud Infrastructure
• Distributed cluster architecture for high performance
• Vector operations executed on HeatWave cluster
• Integrates with OCI Object Storage for document loading
• Requires OCI config for cloud features

​

Prerequisites

• Python 3.8+ with uv package manager
• Database connection file: JSON file with connection details
  • Path from MYSQL_MCP_CONFIG environment variable (absolute or relative)
  • Or local_config.json (default)
• OCI configuration file (~/.oci/config) - only required for OCI Object Storage features

​

Quick Start

1. Create a connection configuration file (see Configuration)
2. Install dependencies and run:

uv run oracle.mysql_mcp_server

3. Configure in your MCP client (e.g., Claude Desktop):

{
  "mcpServers": {
    "mysqltools": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/mysql-mcp-server",
        "run",
        "oracle.mysql_mcp_server"
      ]
    }
  }
}

​

Example Use Cases

​

Database Operations

"List all configured database connections"
"Execute 'SELECT * FROM employees' on my connection"

​

ML and GenAI

"Generate a summary of error logs"
"Add embeddings for 'body' column into 'embedding' column in docs table"
"Ask ml_rag: Show me refund policy from the vector store"
"What is the average delay incurred by flights?"

​

Object Storage

"List all compartments in my tenancy"
"Show all buckets in the development compartment"
"List objects in my 'docs-bucket'"

​

Vector Store

"Load all documents with prefix 'manuals/' into schema hr, table product_docs"
"List local files for vector store ingestion"

​

Next Steps