{"data":{"kind":"file","path":"README.md","version_id":"relg0ajy5ogqjlbq7h4bjcsl","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":1497,"modified_at":"2026-08-05T13:11:05.863000","content_hash":"9a8fea5e9caf9616b9b33fe6eaa36a6c27230ffcc43ceb9421ef2e4bc718a6be"},"entries":[],"content":"# SQL Query Builder & Optimizer\n\nEvaluate an LLM's ability to construct, debug, and optimize SQL queries across multiple difficulty levels and database dialects.\n\n## Overview\n\nThis environment tests practical SQL skills through three task categories:\n\n1. **Query Construction** — Given a schema and natural language requirements, write correct SQL queries (SELECT, JOIN, subqueries, CTEs, window functions).\n2. **Query Optimization** — Given a slow or poorly-written query, rewrite it for better performance using indexes, query restructuring, and execution plan awareness.\n3. **Schema Design** — Given requirements, design normalized schemas with appropriate constraints, indexes, and relationships.\n\n## Difficulty Levels\n\n- **Level 0**: Basic SELECT, WHERE, simple JOINs\n- **Level 1**: Aggregation, GROUP BY, HAVING, subqueries\n- **Level 2**: CTEs, window functions, EXISTS/NOT EXISTS\n- **Level 3**: Complex optimization, covering indexes, query plan analysis\n- **Level 4**: Multi-dialect optimization (PostgreSQL, MySQL, SQLite nuances)\n\n## Evaluation\n\n- **Correctness** (0.4): SQL syntax validity and logical correctness\n- **Optimization** (0.3): Performance-aware patterns (indexes, avoiding N+1, proper JOINs)\n- **Robustness** (0.15): Edge case handling (NULLs, empty sets, duplicates)\n- **Style** (0.15): Readable formatting, aliases, comments for complex queries\n\n## Usage\n\n```python\nimport verifiers as vf\nenv = vf.load_environment(\"samma/sql-query-optimizer\")\n```\n\n## License\n\nMIT\n","encoding":"utf-8","truncated":false,"total_bytes":1497},"status":null}