ABSTRACT
Legal contract review faces critical automation challenges due to complex unstructured contract texts and sophisticated legal reasoning requirements, particularly in Chinese legal contexts where traditional manual methods suffer from low efficiency and error-prone outcomes. PEAT-LLM4LCR addresses these challenges by introducing a comprehensive automated contract review solution specifically optimized for Chinese legal documents, integrating prompt engineering with multi-agent collaboration technology through the Agent-GoTFlow architecture. The system features three core modules for the review of essential elements, the identification of high-risk clauses, and the evaluation of intellectual property, enabling the expression of natural language requirements and generating structured risk reports with modification recommendations. Experimental results across multiple levels of expertise demonstrate improved efficiency and greater accuracy 34% compared to traditional methods, democratizing access to enterprise-grade legal analysis capabilities for legal professionals, business stakeholders and small-medium enterprises with limited legal resources. The tool is available at https://github.com/Hongbin-Xiao/PEAT-LLM4LCR-TOOL, with demonstration video at https://www.youtube.com/watch?v=eyIikQDkv0E.
Xiao, Hongbin and Tan, Shuru and Liu, Wanglong and Li, Ye and Zhou, Xin and Li, Zhi and Xie, Xiaolan, PEAT-LLM4LCR:Chinese Legal Contract Automatic Review Tool Integrating Prompt Engineering and Agent Technology. Posted to SSRN 18 September 2025.
Leave a Reply