Experiment

AI Coding Agent Evaluation

A systematic view of correctness, maintainability, effort and autonomy.

We investigate evaluation designs for coding agents that consider correctness, software quality, maintainability, developer effort, productivity, reliability, context requirements and autonomy. Comparisons must account for task and repository conditions. This entry describes an investigation, not a published benchmark or a completed experiment.

EvaluationEmpirical researchAI4SE