TY  - RPRT
TI  - Can Agent Benchmarks Support Their Scores? Evidence-Supported Bounds for Interactive-Agent Evaluation
AU  - Shanshan Gao
AU  - Liyi Zhou
PY  - 2026
UR  - https://arxiv.org/abs/2605.10448
ID  - 2605.10448
ER  -