GoSQLX

What ajitpratap0/GoSQLX does

ajitpratap0/GoSQLX brings GoSQLX’s SQL tooling into MCP-compatible AI clients. It operates on SQL text rather than querying a database, so it fits code review, query inspection, formatting, validation, and static analysis workflows.

The server provides seven focused operations. Queries can be checked for syntax, converted into a readable layout, summarized as an abstract syntax tree, inspected for referenced objects, reviewed against style rules, scanned for common injection patterns, or processed through a combined analysis. The combined operation runs validation, parsing, metadata extraction, security checks, linting, and formatting concurrently and returns one report.

How it works

Clients send SQL requests to the Streamable HTTP MCP endpoint at https://mcp.gosqlx.dev/mcp. Results are returned as structured data. For example, validation reports whether a statement is valid and can include an error and dialect; metadata extraction returns referenced tables, columns, and functions; parsing returns statement counts and statement types.

Dialect-aware validation is available for PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, SQLite, Snowflake, and ClickHouse. GoSQLX also documents generic, BigQuery, and Redshift dialect names for its broader library and CLI interfaces, although the MCP tool description specifically identifies the eight dedicated dialects above.

The formatter places clauses on separate lines and can apply uppercase keywords and a trailing semicolon. It removes comments from formatted output. Security analysis checks for patterns such as tautologies, UNION-based attacks, stacked queries, and comment-based bypasses. Linting reports violations from rules covering whitespace, indentation, line length, keyword case, comma placement, alias consistency, and related style issues.

Setup and configuration

No local installation is required for the public remote server. Register the endpoint with Claude Code using:

claude mcp add --transport http gosqlx https://mcp.gosqlx.dev/mcp

The same remote URL can be used by Cursor or another MCP client that supports Streamable HTTP. The project also documents adding the URL as a connector in ChatGPT developer mode, where compatible MCP Apps can render the combined analysis as an interactive report.

Tools and capabilities

• validate_sql: Check SQL syntax, optionally for a selected dialect.
• format_sql: Reformat SQL with configurable keyword casing and semicolon output.
• parse_sql: Return an AST-oriented summary of statement count and statement types.
• extract_metadata: Identify referenced tables, columns, and functions.
• lint_sql: Report SQL style violations and their count.
• security_scan: Find common SQL injection patterns and provide severity counts.
• analyze_sql: Produce a composite report from all six analysis stages.

Limitations and notes

The formatter does not preserve comments. The security scanner focuses on detectable SQL injection patterns and should be treated as static analysis rather than a complete application security review. The MCP interface analyzes supplied SQL; it is not presented as a database connection or query execution service. The combined analysis is most useful when a client can display structured MCP results or MCP App content.