What problem does it solve?
Database query, planner, and execution engine changes often introduce subtle behavioral correctness issues and regression risks that are easy to miss during standard pull request reviews, leading to incorrect query results, semantic corruption, or production outages when merged.
Core Features & Use Cases
- Static Analysis-First Review: Performs diff-focused static analysis of code changes without running build or test commands, keeping review latency low.
- Comprehensive Correctness Checklist: Runs a mandatory, section-by-section checklist covering SQL semantics, parser/binder rules, optimizer logic, execution operator contracts, and behavior-preservation requirements.
- High-Risk Edge Case Gates: Includes specialized validation gates for high-risk areas like stream tokenization, chunked record framing, reverse/forward scan paths, and pushdown/rewrite operator preservation to catch boundary-related regressions.
- Use Case: A database engineering team merging a query optimizer change can use this skill to verify that predicate pushdown logic correctly handles NULL values and outer-join semantics, and that new execution operators do not drop required rows in edge cases.
Quick Start
Use the review-correctness-query-planner-execution skill to review the pull request diff for query, planner, and execution engine changes in your cloned database codebase, outputting structured correctness findings to the specified JSON file.