Draft suggestion
SQL Optimization Workbench
Move from slow SQL to a reviewable optimization report.
Paste a business problem, slow query, schema, EXPLAIN, collector output, or benchmark evidence. The project flow turns it into findings, proof steps, and a report.
AI-generated candidates. Review and test before production use.
Evidence levels
Bring what you have. Improve the result as evidence grows.
The system labels the output by evidence level, so a rough idea does not look like a verified production result.
Draft query review
Candidate index or rewrite
Evidence-backed finding
Verified report section
Candidate SQL
Index and rewrite candidates stay compact, with a review note instead of a wall of code.
Validation steps
EXPLAIN, benchmark, and manual checks are listed as next actions inside the project.
Safety boundary
Destructive SQL gets a visible warning. The app does not auto-apply production changes.