AI coding agents make empirical specification search fast and cheap, but they also widen hidden researcher degrees of freedom. Building on an open-source agent-loop architecture, this paper recasts a minimal coding loop as a transparent protocol for empirical economics. In a forecast-combination illustration, multiple independent agent runs outperform standard benchmarks in the original rolling evaluation, but not all continue to do so on a post-search holdout. Logged search and holdout evaluation together make adaptive specification search visible and help distinguish robust improvements from sample-specific discoveries.

翻译：AI编码代理使得实证设定搜索变得快速且廉价，但也扩大了隐藏的研究者自由度。本文基于一种开源代理循环架构，将最小化编码循环重新构建为实证经济学的透明协议。在一个预测组合的示例中，多个独立代理运行在原始滚动评估中优于标准基准，但并非所有代理都能在搜索后保留样本上持续保持优势。记录的搜索过程与保留样本评估共同使自适应设定搜索变得可见，并有助于区分稳健改进与样本特异性发现。