In order to meet the needs of students' programming debugging ability training, this paper designs and implements a data acquisition and analysis system for programming debugging process based on VS Code plug-in, which aims to solve the limitation of traditional assessment methods that are difficult to fully evaluate students' debugging ability. The system supports a variety of programming languages, integrates debugging tasks and data acquisition functions, captures students' debugging behavior in the local editor in real time, and uploads the data to the platform database to realize the whole process monitoring and feedback, provides accurate debugging guidance for teachers, and improves the teaching effect. In terms of data analysis, the system proposed a debugging behavior analysis model based on abstract syntax tree, combined with node annotation, sequence recognition and cluster analysis and other technologies, to automatically track the context of students' debugging process and accurately identify key features in the debugging path. Through this tool, the system realizes the intelligent identification and labeling of the debugging direction and behavior pattern, and improves the refinement level of debugging data analysis. In this research system, a complex debugging scenario of multi-file and multi-task is introduced into the debugging problem design, which optimizes the multi-dimensional capturing ability of debugging data and lays a foundation for accurate debugging behavior analysis. Through several practical teaching tests, the feasibility and stability of the system are verified, which proves that it can effectively support procedural evaluation in programming debugging teaching, and provides a new direction for debugging behavior analysis research.