AdaCLIP is a novel method for zero-shot anomaly detection (ZSAD) that leverages a pre-trained vision-language model (VLM), CLIP, and introduces hybrid learnable prompts to enhance ZSAD performance. The method incorporates both static and dynamic prompts, which are used to adapt CLIP for ZSAD. Static prompts are shared across all images, while dynamic prompts are generated for each test image. The combination of static and dynamic prompts, referred to as hybrid prompts, significantly improves ZSAD performance. AdaCLIP is evaluated on 14 real-world anomaly detection datasets from industrial and medical domains, demonstrating superior performance compared to other ZSAD methods. The method also shows strong generalization capabilities across different categories and domains. The analysis highlights the importance of diverse auxiliary data and optimized prompts for enhanced generalization. The code for AdaCLIP is available at https://github.com/caoyunkang/AdaCLIP. Keywords: Anomaly Detection · Prompt Learning · Zero-shot Learning.AdaCLIP is a novel method for zero-shot anomaly detection (ZSAD) that leverages a pre-trained vision-language model (VLM), CLIP, and introduces hybrid learnable prompts to enhance ZSAD performance. The method incorporates both static and dynamic prompts, which are used to adapt CLIP for ZSAD. Static prompts are shared across all images, while dynamic prompts are generated for each test image. The combination of static and dynamic prompts, referred to as hybrid prompts, significantly improves ZSAD performance. AdaCLIP is evaluated on 14 real-world anomaly detection datasets from industrial and medical domains, demonstrating superior performance compared to other ZSAD methods. The method also shows strong generalization capabilities across different categories and domains. The analysis highlights the importance of diverse auxiliary data and optimized prompts for enhanced generalization. The code for AdaCLIP is available at https://github.com/caoyunkang/AdaCLIP. Keywords: Anomaly Detection · Prompt Learning · Zero-shot Learning.