db-schema-map: Schema Map from MyBatis & JPA

Reverse-engineering a legacy database schema from a large Spring codebase is slow, especially without a live database connection. This skill builds a table-centric schema map by statically analyzing MyBatis XML mappers and JPA/Hibernate entities. It extracts table relationships, column definitions, and CRUD access patterns directly from the source code.

What it does

• ORM auto-detection identifies MyBatis or JPA/Hibernate configurations to apply the correct parsing logic.
• Dependency mapping resolves table relationships and N:N join tables from @JoinTable or MyBatis ON clauses.
• Impact analysis tracks call sites for specific tables to determine the blast radius of a schema change.
• Risk signaling flags blind spots like SQL injection risks in ${} substitutions or hidden QueryDSL queries.
• CRUD profiling categorizes table usage by SELECT, INSERT, UPDATE, and DELETE operations.

How it works

1. Run the script against a project root to generate a full domain map.
2. Target a specific table name to see every DAO method or Repository call touching that data.
3. Export results to JSON for machine-readable integration with other internal tooling.
4. Review the Signals section to gauge code quality and naming strategy confidence.

Frameworks & tools

Python 3.8+ standard library. Works with Spring Boot, MyBatis, JPA, and Hibernate. Supports Flyway and Liquibase detection.

Why this beats prompting it yourself

Manual prompts struggle to maintain state across hundreds of XML and Java files, often missing aliased joins or dynamic SQL tags. This tool uses structured regex scanning to ensure every DAO.statementId and Repository method is accounted for without losing context.

Use cases

• Onboarding to an unfamiliar project by identifying the top 5 core domain tables.
• Pre-migration audits to find all code paths affected by renaming a column or table.
• Security reviews to locate every instance of unsafe string substitution in MyBatis.
• Performance tuning by identifying EAGER fetch types and N+1 risks in JPA entities.

Known limitations

Uses regex scanning rather than full SQL/Java parsing. Accuracy is not compiler-grade. MyBatis column extraction is limited to write paths. Does not detect QueryDSL or Criteria API query logic.