CRISPR-GPT is an LLM agent designed to automate and enhance the design of CRISPR-based gene-editing experiments. It integrates domain knowledge and external tools to assist researchers in selecting CRISPR systems, designing guide RNAs, recommending delivery methods, drafting protocols, and validating editing outcomes. The agent supports various tasks, including Meta Mode for predefined pipelines, Auto Mode for task decomposition, and Q&A Mode for answering queries. CRISPR-GPT was evaluated by 12 experts, who found it significantly more accurate and effective than general LLMs in gene-editing design tasks. It was also validated through a real-world experiment where it successfully facilitated gene knockout in human cells. CRISPR-GPT ensures ethical and safe use by restricting human genome editing and protecting user data privacy. The system demonstrates the potential of LLM agents in complex biological research, enabling efficient and accurate gene-editing experiments. However, it has limitations, such as the inability to generate complete constructs from natural language input, and future improvements are needed. The integration of CRISPR-GPT with automated laboratory platforms could further enhance its capabilities in accelerating scientific discoveries.CRISPR-GPT is an LLM agent designed to automate and enhance the design of CRISPR-based gene-editing experiments. It integrates domain knowledge and external tools to assist researchers in selecting CRISPR systems, designing guide RNAs, recommending delivery methods, drafting protocols, and validating editing outcomes. The agent supports various tasks, including Meta Mode for predefined pipelines, Auto Mode for task decomposition, and Q&A Mode for answering queries. CRISPR-GPT was evaluated by 12 experts, who found it significantly more accurate and effective than general LLMs in gene-editing design tasks. It was also validated through a real-world experiment where it successfully facilitated gene knockout in human cells. CRISPR-GPT ensures ethical and safe use by restricting human genome editing and protecting user data privacy. The system demonstrates the potential of LLM agents in complex biological research, enabling efficient and accurate gene-editing experiments. However, it has limitations, such as the inability to generate complete constructs from natural language input, and future improvements are needed. The integration of CRISPR-GPT with automated laboratory platforms could further enhance its capabilities in accelerating scientific discoveries.