Legacy SQL Architect MCP

What it is

A Java server for PostgreSQL, SQL Server and Oracle that connects strictly read-only. It retrieves full stored-procedure source, traces dependencies between tables, triggers, procedures and views, analyses execution plans, and generates documentation and JDBC code from the live schema.

What you get

• inspect_schema — tables, columns, types, primary keys and foreign keys
• get_procedure_source — the complete source of stored procedures and functions, including large PL/SQL packages
• data_sampler — sample rows with sensitive columns masked automatically
• query_plan_expert — execution plan analysis with full table scan detection
• dependency_graph and find_impact — foreign key chains, trigger to procedure call chains, view dependencies, and everything that depends on a given table
• generate_mermaid_erd and generate_documentation — an ER diagram and full Markdown docs for the schema
• generate_java_dao — Entity and Repository classes in plain JDBC per table
• An Oracle APEX toolkit: inspect_apex_performance, get_apex_source, inspect_apex_debug, apex_config_audit, apex_sql_runtime_stats and apex_explain_batch

Requirements

A native installer on Windows, Debian or RHEL, or the JAR with Java 21 or newer. Connection details go in a config.json in the install directory: database type, JDBC URL, user and password. Tested against PostgreSQL 11+, SQL Server 2016+ and Oracle 12c+. No Docker, no ORM.

Setup effort

Follow the published steps

Tools you get

15 tools named in the project's documentation. Six of the 15 are Oracle APEX tools that work only on Oracle, so PostgreSQL and SQL Server users get 9. apex_sql_runtime_stats needs SELECT on V$SQL or returns a warning.