PHP Coding-Agent Benchmark

Engineering work by Orçun Candan.

Implementation

Task-based evaluator for PHP coding agents covering snapshot isolation, idempotency, persistence, concurrency and crash recovery, with static analysis and mutation scoring.

Results & scope

Five task variants with recorded runs and calibration controls. Small run counts limit comparisons between closely scoring models.

Tools & technologies

  • PHP
  • Python
  • PHPStan
  • Mutation testing
  • Agent evaluation

Sources & public evidence

All projects · Contact Orçun Candan