Skip to main content
Glama
mcp-mysql-explorer

mcp-mysql-explorer

by xsg22

MCP MySQL Server

A Model Context Protocol (MCP) server that connects AI agents (Cursor, Claude Desktop, etc.) to MySQL databases, enabling schema exploration, data querying, and SQL execution through natural language.

Chinese docs: README.zh.md

Features

Tool

Description

list_tables

List all tables in the database

describe_table

View table schema with column types and comments

query

Execute read-only queries (SELECT / SHOW / DESCRIBE / EXPLAIN)

execute_sql

Execute write operations (INSERT / UPDATE / DELETE / DDL), disabled in read-only mode

get_table_indexes

Show all indexes of a table

get_create_table

Get the CREATE TABLE statement

get_database_info

Get database overview (version, size, table count)

Related MCP server: DatabaseMCP

Installation

pip install mcp-mysql-explorer

Usage in Cursor

Open Cursor Settings -> MCP, add a new server. Three ways to configure the database connection. SSH tunneling through a bastion/jump host is also supported.

{
  "mcpServers": {
    "mysql": {
      "command": "mcp-mysql-explorer",
      "args": [
        "--host", "your-mysql-host",
        "--port", "3306",
        "--user", "your-user",
        "--password", "your-password",
        "--database", "your-database"
      ]
    }
  }
}

Enable write mode explicitly:

{
  "mcpServers": {
    "mysql": {
      "command": "mcp-mysql-explorer",
      "args": [
        "--host", "your-mysql-host",
        "--database", "your-database",
        "--allow-write"
      ]
    }
  }
}

If the database is only reachable through an SSH bastion, add --ssh-* arguments:

{
  "mcpServers": {
    "mysql": {
      "command": "mcp-mysql-explorer",
      "args": [
        "--host", "mysql.internal",
        "--port", "3306",
        "--user", "your-db-user",
        "--password", "your-db-password",
        "--database", "your-database",
        "--ssh-host", "bastion.example.com",
        "--ssh-port", "22",
        "--ssh-user", "your-ssh-user",
        "--ssh-private-key", "/Users/you/.ssh/id_rsa"
      ]
    }
  }
}

In SSH mode, --host / MYSQL_HOST is the MySQL address as seen from the SSH server. SSH tunneling is enabled automatically when --ssh-host or MYSQL_SSH_HOST is set. You can explicitly disable it with --ssh-enabled false.

Option 2: Environment variables

{
  "mcpServers": {
    "mysql": {
      "command": "mcp-mysql-explorer",
      "env": {
        "MYSQL_HOST": "your-mysql-host",
        "MYSQL_PORT": "3306",
        "MYSQL_USER": "your-user",
        "MYSQL_PASSWORD": "your-password",
        "MYSQL_DATABASE": "your-database",
        "MYSQL_READ_ONLY": "true",
        "MYSQL_SSH_HOST": "bastion.example.com",
        "MYSQL_SSH_PORT": "22",
        "MYSQL_SSH_USER": "your-ssh-user",
        "MYSQL_SSH_PRIVATE_KEY": "/Users/you/.ssh/id_rsa"
      }
    }
  }
}

SSH environment variables:

Environment variable

Description

MYSQL_SSH_ENABLED

Enable SSH tunneling (true/false). Defaults to enabled when MYSQL_SSH_HOST is set

MYSQL_SSH_HOST

SSH bastion/jump host

MYSQL_SSH_PORT

SSH port, defaults to 22

MYSQL_SSH_USER

SSH username

MYSQL_SSH_PASSWORD

SSH password

MYSQL_SSH_PRIVATE_KEY

Path to SSH private key

MYSQL_SSH_PRIVATE_KEY_PASSWORD

SSH private key passphrase

MYSQL_SSH_LOCAL_HOST

Local forwarding bind host, defaults to 127.0.0.1

MYSQL_SSH_LOCAL_PORT

Local forwarding bind port, defaults to 0 for automatic assignment

Option 3: .env file

Create a .env file in the working directory:

MYSQL_HOST=your-mysql-host
MYSQL_PORT=3306
MYSQL_USER=your-user
MYSQL_PASSWORD=your-password
MYSQL_DATABASE=your-database
MYSQL_READ_ONLY=true
MYSQL_SSH_HOST=bastion.example.com
MYSQL_SSH_USER=your-ssh-user
MYSQL_SSH_PRIVATE_KEY=/Users/you/.ssh/id_rsa

Then simply:

{
  "mcpServers": {
    "mysql": {
      "command": "mcp-mysql-explorer"
    }
  }
}

Priority: command-line args > environment variables > .env file

Usage in Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "mysql": {
      "command": "mcp-mysql-explorer",
      "args": [
        "--host", "your-mysql-host",
        "--user", "your-user",
        "--password", "your-password",
        "--database", "your-database"
      ]
    }
  }
}

Run Manually

# With command-line arguments
mcp-mysql-explorer --host localhost --user root --password secret --database mydb

# Explicitly allow write operations
mcp-mysql-explorer --host localhost --user root --password secret --database mydb --allow-write

# Through an SSH bastion
mcp-mysql-explorer --host mysql.internal --user root --password secret --database mydb \
  --ssh-host bastion.example.com --ssh-user deploy --ssh-private-key ~/.ssh/id_rsa

# With .env file
mcp-mysql-explorer

# As a Python module
python -m mcp_mysql_explorer --host localhost --database mydb

Testing

# Run unit tests (no real database needed)
python scripts/run_tests.py
# Optional manual smoke check (reads credentials from env only)
# Required env: MYSQL_HOST, MYSQL_PORT, MYSQL_USER, MYSQL_PASSWORD, MYSQL_DATABASE
python scripts/manual_smoke_check.py --mode both

Security Notes

  • The query tool only allows SELECT / SHOW / DESCRIBE / EXPLAIN statements.

  • Read-only mode is enabled by default (MYSQL_READ_ONLY=true).

  • To enable writes, set MYSQL_READ_ONLY=false or pass --allow-write.

  • The execute_sql tool can run write operations only when read-only mode is disabled.

  • Query results are capped at 1000 rows by default.

  • Table/column identifiers are backtick-wrapped to prevent SQL injection.

License

MIT

Available Tools

7 tools
describe_tableC

Get the schema (columns, types, keys) of a specific table.

Args: table_name: Name of the table to describe.

Inputs: table_name (required). Returns: result

execute_sqlB

Execute a write SQL statement (INSERT / UPDATE / DELETE / ALTER / CREATE / DROP, etc).

Use with caution: this modifies the database.

Args: sql: The SQL statement to execute.

Inputs: sql (required). Returns: result

get_create_tableC

Get the CREATE TABLE statement of a specific table.

Args: table_name: Name of the table.

Inputs: table_name (required). Returns: result

get_database_infoA

Get overview information about the connected database, including size, table count, and charset.

Inputs: no named parameters. Returns: result

get_table_indexesB

Show all indexes of a specific table.

Args: table_name: Name of the table.

Inputs: table_name (required). Returns: result

list_tablesA

List all tables in the connected MySQL database.

Inputs: no named parameters. Returns: result

queryA

Execute a read-only SQL query (SELECT / SHOW / DESCRIBE / EXPLAIN).

Only SELECT, SHOW, DESCRIBE, and EXPLAIN statements are allowed.

Args: sql: The SQL query to execute.

Inputs: sql (required). Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.3/5.0

    Scored across 7 tools

    Disambiguation4/5

    The read/write split between query and execute_sql is explicitly stated in both descriptions, so an agent can pick correctly. The introspection tools (describe_table, get_table_indexes, get_create_table) overlap somewhat since DESCRIBE/SHOW can also surface keys and index info, but each has a distinct output focus.

    Naming Consistency3/5

    All names are snake_case, but verb conventions are mixed: list_tables/describe_table use one verb family, while get_table_indexes/get_create_table/get_database_info use 'get', and query/execute_sql are bare verb forms. Readable overall, but 'get_create_table' is awkward and the pattern isn't uniform.

    Tool Count5/5

    Seven tools is well-scoped for a MySQL explorer: schema discovery, ad-hoc read queries, writes, and database metadata each have exactly one owner. No tool feels redundant padding.

    Completeness4/5

    The surface covers introspection, read queries, write DDL/DML, and database-level info, so core workflows are supported. Minor gaps exist around database/schema switching (e.g. listing databases or selecting a schema) and explicit transaction control, which agents would have to work around with raw SQL.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      C
      maintenance
      An MCP server for MySQL databases that enables schema exploration, query execution, and resource access through a unified interface. It features a LangGraph-based agent that translates natural language into SQL queries with automatic error recovery and schema discovery.
      -
    • F
      license
      Not graded
      quality
      C
      maintenance
      An MCP server that enables AI agents to safely explore and interact with MySQL databases through dynamic tool generation from stored procedures. It provides database discovery capabilities and intelligent procedure categorization while enforcing security restrictions to prevent data modification.
      -
    • F
      license
      Not graded
      quality
      B
      maintenance
      An MCP server that exposes relational databases (PostgreSQL/MySQL) to AI agents with natural language to SQL query support.
      19
      -
    • A
      license
      Not graded
      quality
      D
      maintenance
      A MySQL MCP server for secure database interaction, enabling schema inspection, query execution, and RBAC via AI coding assistants.
      7
      MIT