Vision–language foundation models such as CLIP are increasingly used for zero-shot behavior recognition, yet their robustness to prompt variations remains poorly understood. This paper investigates prompt sensitivity as a critical robustness concern in CLIP-family models for zero-shot classroom behavior analysis, treating prompt wording as a controlled semantic perturbation. Five representative vision–language models (CLIP/OpenAI, OpenCLIP/LAION, SigLIP2, EVA02-CLIP, and DFN-CLIP) are evaluated on three public classroom behavior benchmarks under a strict symmetric protocol. We compare four generic prompt strategies with a training-free Class-Aware Prompt Ensemble (CAPE). Results show that minor prompt changes can cause catastrophic performance degradation. On SigLIP2, an alternative wording of CAPE reduces Hit@1 on TeacherBehavior from 85.5% to 31.4%, a 54.1 percentage-point drop that exceeds the differences between model backbones. Across all five models, action-oriented prompts improve Hit@1 by up to 54 percentage points compared with label-only prompts. We further demonstrate that the apparent superiority of zero-shot CLIP over supervised linear probes largely arises from metric asymmetry. While zero-shot methods achieve higher Hit@1, they consistently underperform linear probes in multi-label evaluation (Sample-F1: 60–66% vs. 88–90%; Macro-F1: 49–60% vs. 68–78%). Bootstrap confidence intervals and paired-bootstrap significance tests further show that several reported performance differences are not statistically significant. These findings reveal prompt sensitivity as a fundamental deployment risk for vision–language foundation models in domain-specific behavior analysis. Prompt variations involving only a few words can silently undermine recognition performance while remaining hidden by conventional evaluation metrics. We therefore recommend that future benchmark studies report prompt configurations, multi-label F1 scores, and uncertainty estimates alongside headline Hit@1 to provide a more complete and reliable assessment of model capability.

The construction of soil surface microtopography not only effectively mitigates soil erosion, improves soil structure, and enhances soil ecological functions, but also significantly optimizes the seedbed environment for seedling emergence and crop growth. In this study, targeting the specific characteristics of red-yellow soils in Southern China, a quadrangular frustum-shaped soil surface microtopography processing device was designed and fabricated based on the 2BYG-230 rapeseed seeder. The motion trajectory and force distribution of the device were analyzed using the Discrete Element Method (DEM) software, EDEM, followed by three-factor and three-level orthogonal tests. The results indicated that the order of significance for factors affecting the microtopography formation effect was working load > working speed > microstructure height. Using the formation qualification rate as the evaluation index, the soil disturbance patterns were analyzed to determine the optimal combination of operating parameters: a working load of 260 N, a working speed of 0.34 m/s, and a microstructure height of 42 mm. Under these optimized conditions, the microtopography formation qualification rate reached 93.6%. Furthermore, the seedling emergence rate following the operation of the optimized device was 74.33%, representing a 4.96% increase compared to pre-optimization levels. The optimized processing device designed in this study markedly outperformed its predecessor, creating a soil surface microtopography more conducive to rapeseed growth while demonstrating substantial potential for water and soil conservation and ecological improvement. This research provides theoretical support for enhancing the ecological functions of Southern red-yellow soils and for the structural design of surface microtopography processing equipment.