Skip to directory

Oracle MySQL MCP Server (POC)

Oracle (proof-of-concept project)

stdio14 documented tools1

About Oracle MySQL MCP Server (POC)

Oracle-published proof-of-concept MCP server for selected MySQL and HeatWave workflows. The upstream project explicitly says it is not intended for production use.

What you can use it for

  • Explore documented MySQL and HeatWave MCP workflows in a development environment
Connection and setup
Hosting
Local process
Transport
stdio
Endpoint
Local process; no hosted endpoint documented
Authentication
MySQL connection settings are supplied through the documented config file. OCI credentials are required for HeatWave/Object Storage-backed modes; MySQL AI mode has different prerequisites.

Requirements

  • Python and uvx
  • MySQL server for the selected mode
  • A local configuration file identified through MYSQL_MCP_CONFIG
  • Oracle Cloud Infrastructure credentials/config only for modes that use OCI services

Setup steps

  1. Review the upstream POC warning and choose the MySQL mode you intend to use before installing.
  2. Create the MySQL MCP configuration file following the repository README and set MYSQL_MCP_CONFIG to its local path in the MCP client environment.
  3. For a mode requiring OCI, configure OCI credentials through Oracle’s documented local config mechanism; do not put credentials in this catalog or prompt.
  4. Run `uvx oracle.mysql-mcp-server` as the stdio command.
  5. Reconnect and inspect the discovered tools; tool availability depends on configured mode. Restrict database permissions and do not treat this POC as production-ready.
Tools 14

The current Oracle README enumerates 14 API tools. This is a proof of concept and is not intended for production use.

  • list_all_connections

    List configured database connections and modes.

    01
  • execute_sql_tool_by_connection_id

    Execute SQL on a configured MySQL connection.

    02
  • ml_generate

    Generate text using MySQL AI or HeatWave.

    03
  • ragify_column

    Create and populate vector embeddings for a column.

    04
  • list_vector_store_files_local

    List files available in secure_file_priv.

    05
  • load_vector_store_local

    Load local documents into a vector store.

    06
  • load_vector_store_oci

    Load documents from OCI Object Storage into a vector store.

    07
  • ask_ml_rag_vector_store

    Query the default vector store with retrieval-augmented generation.

    08
  • ask_ml_rag_innodb

    Query InnoDB vector segments with retrieval-augmented generation.

    09
  • heatwave_ask_help

    Ask natural-language questions about HeatWave ML.

    10
  • list_all_compartments

    List OCI compartments.

    11
  • object_storage_list_buckets

    List buckets in an OCI compartment.

    12
  • object_storage_list_objects

    List objects in an OCI bucket.

    13
  • ask_nl_sql

    Convert a natural-language question to SQL and execute it.

    14
Limitations and notes
  • Upstream explicitly labels this a proof-of-concept and says it is not intended for production use.
  • Tool availability and credential requirements vary by MySQL/HeatWave mode.
  • Database and cloud operations may expose sensitive data or cause changes; use a disposable or least-privilege development environment.
Original sources

Documentation checked 2026-10-28T00:00:00Z. Connection has not been runtime tested.