What this prompt does
Review a SQL query for correctness, security, performance, and maintainability.
Coding
SQL Query Review
Review a SQL query for correctness, security, performance, and maintainability.
- Models
- ChatGPT, Claude
- Level
- Advanced
- Version
- v0.9
Editorial template. Model evaluation is not recorded; the preview illustrates the intended result. Testing Method
Best for
Correctness issues, injection risks, index considerations, rewritten query, EXPLAIN plan checklist, and edge cases.
Prompt variables
[SQL][SCHEMA][DATABASE][GOAL]System Prompt
You are a database reviewer. Treat schema and execution-plan facts as unknown unless supplied.
User Prompt
OBJECTIVE: Review [SQL] against [SCHEMA] on [DATABASE] for the goal [GOAL]. INPUTS: SQL: [SQL] SCHEMA: [SCHEMA] DATABASE: [DATABASE] GOAL: [GOAL] OUTPUT CONTRACT: Correctness issues, injection risks, index considerations, rewritten query, EXPLAIN plan checklist, and edge cases.
How to use
State the business meaning of the result and provide a tiny fixture with expected rows. Review joins and filters for correctness before discussing performance. Share existing indexes and an execution plan if performance is part of the task. Ask the assistant to distinguish a logical error from an index hypothesis and to preserve the database’s actual behavior.
Why it works
The blueprint separates authority, task context, evidence limits, and the output contract so requirements can be checked independently.
Model compatibility
05
Example input
SQL: SELECT customer_id, COUNT(*) FROM orders GROUP BY customer_id SCHEMA: orders(id, customer_id, status) DATABASE: MySQL; supply actual version for a real review GOAL: Count completed orders per customer
06
Example output
Correctness finding: The query counts all statuses, not completed orders. Candidate query: SELECT customer_id, COUNT(*) FROM orders WHERE status = 'completed' GROUP BY customer_id; Verification: Use one customer with two completed orders and one canceled order; expected count is two. Performance: Inspect EXPLAIN and representative table sizes before recommending an index. No speedup is measured here.
How to customize
For a join, include one unmatched row and one relationship with multiple matches. For date filtering, specify the column type and timezone boundary. Keep any proposed index separate from the query correction so you can validate each change independently.
Common mistakes
Do not claim a performance improvement without measuring it. A faster query that counts the wrong rows is not a successful optimization.
Test notes
Editorial blueprint awaiting recorded cross-model evaluation. It is visible for review but is not labeled as tested.
Why is this blueprint not marked as tested?
It has passed editorial structure review but does not yet have a published cross-model test record.
07
Related prompts
Coding
Senior Code Review
Coding
PHP Debugging Investigation
Marketing
Customer Acquisition Blueprint
SEO
Technical SEO Audit
Writing
Editorial Content Brief
Research