MCP Alchemy

Content

MCP Alchemy is a SQLAlchemy-backed Model Context Protocol server that connects Claude to relational databases. It can list table names, filter tables, inspect schemas and foreign-key relationships, and execute SQL queries against a configured database URL.

The project supports SQLAlchemy-compatible databases, including PostgreSQL, MySQL, MariaDB, SQLite, Oracle, Microsoft SQL Server, CrateDB, Vertica, and other dialects when the matching Python driver is installed.

Source Review

• https://github.com/runekaagaard/mcp-alchemy
• https://github.com/runekaagaard/mcp-alchemy/blob/main/README.md
• https://pypi.org/pypi/mcp-alchemy/json
• https://github.com/runekaagaard/mcp-alchemy/blob/main/pyproject.toml
• https://github.com/runekaagaard/mcp-alchemy/blob/main/mcp_alchemy/server.py
• https://github.com/runekaagaard/mcp-alchemy/blob/main/LICENSE

These sources were reviewed on 2026-06-06. Prefer the live repository, README, PyPI metadata, package metadata, server implementation, and license for current install commands, package versions, supported databases, environment variables, tool behavior, and licensing.

Features

• PyPI package mcp-alchemy.
• SQLAlchemy-based database connectivity.
• Support for PostgreSQL, MySQL, MariaDB, SQLite, Oracle, MS SQL Server, CrateDB, Vertica, and other SQLAlchemy dialects.
• all_table_names tool for listing database tables.
• filter_table_names tool for matching tables by substring.
• schema_definitions tool for columns, primary keys, nullable flags, and foreign-key relationships.
• execute_query tool for SQL execution with optional parameters.
• EXECUTE_QUERY_MAX_CHARS output truncation.
• Optional DB_ENGINE_OPTIONS for SQLAlchemy engine settings.
• Optional CLAUDE_LOCAL_FILES_PATH integration for full result sets.
• Mozilla Public License 2.0 licensing.

Installation

Configure MCP Alchemy with the database driver required by the target database. For example, start with the generic package and add the appropriate --with driver from the README when needed:

{
  "mcpServers": {
    "mcp-alchemy": {
      "command": "uvx",
      "args": ["--from", "mcp-alchemy", "mcp-alchemy"],
      "env": {
        "DB_URL": "<sqlalchemy-database-url>",
        "EXECUTE_QUERY_MAX_CHARS": "4000"
      }
    }
  }
}

Use a dedicated read-only database user for exploration workflows. Add write privileges only when the workflow is reviewed and intended to modify data.

Use Cases

• Ask Claude to list tables before exploring an unfamiliar relational database.
• Inspect schemas and foreign-key relationships before drafting SQL.
• Run reviewed SQL queries against a development or analytics database.
• Generate reports from query results with controlled truncation.
• Connect to different database engines using SQLAlchemy dialects.
• Use Claude Local Files for full result sets only when local result export is approved.

Safety and Privacy

MCP Alchemy gives Claude a generic SQL execution surface. It does not enforce read-only behavior, and the source configures AUTOCOMMIT, so database permissions are the real guardrail. Use least-privilege credentials, prefer read-only users, and review every query before execution.

Connection URLs and query results can expose sensitive database details. Avoid production credentials, regulated datasets, broad admin users, and exported full-result files unless the model session and storage location are approved for that data.

Duplicate Check

Existing content includes DBHub, MariaDB MCP Server, DuckDB MCP Server, StarRocks MCP Server, and other database-specific MCP entries. This entry is distinct because it covers runekaagaard/mcp-alchemy, a SQLAlchemy-compatible multi-database MCP server with schema relationship inspection, vertical query output, and optional Claude Local Files integration. No matching source URL or dedicated MCP Alchemy entry was found in content/mcp.