mcp -sql-analyzer

This MCP server provides SQL analysis capabilities, including syntax linting, dialect conversion, and extraction of table/column references. Clients can use the provided tools to validate SQL, transpile between dialects (for example, MySQL to PostgreSQL), and inspect dependencies and references within a query. The server exposes four primary tools: lint_sql for syntax validation, transpile_sql for dialect conversion, get_all_table_references to enumerate table and CTE usage, and get_all_column_references to enumerate column references with their table context. To leverage these, connect via MCP protocol to the sql-analyzer server name and invoke the desired tool with the appropriate input SQL and optional dialects. This enables automated SQL verification and analysis within larger workflows or AI-assisted SQL authoring environments.

Prerequisites:

• Node.js installed (for general MCP tooling compatibility) or a Python environment capable of running UV/uvx as per your setup
• Access to the internet to install dependencies

Installation steps (recommended approach using uvx):

1. Ensure uvx is installed (Python package runner used by MCP):
   • With Python 3.8+ and pip available, install uvx via your environment's package manager or pip if applicable:
     • pip install uvx (if available for your setup)
   2. Run the MCP server using the provided configuration (from the README example): uvx --from git+https://github.com/j4c0bs/mcp-server-sql-analyzer.git mcp-server-sql-analyzer

Alternative (if you prefer using uv):

1. Clone the repository locally: git clone https://github.com/j4c0bs/mcp-server-sql-analyzer.git
2. Run the server from the cloned directory using uv: uv --directory /path/to/mcp-server-sql-analyzer run mcp-server-sql-analyzer

Verification:

• Ensure the MCP client can connect to the sql-analyzer server name defined in your claude_desktop_config.json or equivalent config file.
• Validate that lint_sql, transpile_sql, get_all_table_references, and get_all_column_references respond with expected results for sample queries.

Tips and common considerations:

• When using lint_sql, provide the target dialect if known to improve accuracy (e.g., 'mysql' or 'postgres').
• For transpile_sql, ensure both read_dialect and write_dialect are specified to receive a valid transpiled query.
• The get_all_table_references and get_all_column_references tools return structured results including tables, aliases, and fully qualified names; these are useful for dependency analysis and query comprehension.
• If you encounter environment issues, verify that your uvx/uv installation is compatible with your Python environment and that network access to fetch the git repository is enabled.
• Dialect discovery can be queried via the dialects://all endpoint to enumerate supported dialects for use with the tools.