Evaluate LLM SQL Generation
Independent PiSkill directory guide. The original prompt remains hosted by OpenAI Cookbook.
What does this prompt do?
Builds an evaluation workflow for generated SQL using representative cases, execution-aware checks, correctness criteria, and reproducible comparison rather than subjective inspection alone.
Primary use case
Evaluate whether model-generated SQL is correct and useful.
Expected output
Repeatable SQL-generation evaluation results.
Inputs or variables
- SQL tasks
- database schema
- expected behavior
Related prompts
Claude SQL Query Generation
Demonstrates a practical SQL-generation workflow that supplies schema context, asks for the correct dialect, constrains changes, and checks generated queries before execution.
Claude Text-to-SQL
Generates SQL from a natural-language question using explicit schema context, dialect constraints, validation, and handling for ambiguous or unsafe requests.
Autofix GitHub Actions with Codex
Demonstrates a Codex workflow that reads failed GitHub Actions evidence, identifies the responsible change or configuration, proposes a focused repair, and validates it.
BigQuery AI Operators
Demonstrates BigQuery AI operators for applying model capabilities to warehouse data with explicit SQL, input columns, result handling, and cost-aware validation.
BigQuery RAG with Gemini
Builds a retrieval-augmented question-answering workflow with BigQuery data, embeddings, retrieval, prompt context, and grounded Gemini responses.
Codex Prompting Guide
Provides a source-linked recipe for structuring coding tasks, context, constraints, tool use, iteration, and verification when working with Codex-oriented models.