From Idea to Classroom in Days: Using "Vibe Coding" to Create a Programming Process Visualizer from IDE Activity Logs
This paper presents the rapid development and classroom deployment of an AI-assisted "vibe coding" web application that analyzes Thonny IDE logs to generate interactive visualizations of students' programming processes, thereby enabling teachers to efficiently monitor learning progress, identify support needs, and detect potential academic integrity issues.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are a detective trying to solve a mystery, but the only clue you have is the final photo of the crime scene. You can see the result, but you have no idea how the suspect got there. Did they solve it with a brilliant flash of insight, or did they copy the answer from a neighbor and just change the color of the shirt? In the world of computer science, specifically in the classroom, teachers often face this exact problem. They see the final code a student submits, but the messy, trial-and-error journey of writing it remains hidden. This field, known as learning analytics, tries to peek behind the curtain by studying the "activity logs"—digital footprints left behind every time a student types, deletes, runs, or errors out of a program. The big question is: How can we turn these boring, invisible logs into a clear story that helps teachers spot struggling students or catch cheating without spending hours staring at screens?
Enter the "Thonny Log Visualizer," a new tool built by researchers at the University of Tartu in Estonia. Think of this tool as a magical "time-travel camera" for coding. Instead of just showing the final photo, it reconstructs the entire movie of how a student built their program, second by second. The researchers created this tool using a trendy, fast-paced method called "vibe coding," where they acted like directors giving natural-language instructions to an AI assistant to build the software in just a few days. The result is a web application that takes the raw, messy data from a student's coding session and turns it into colorful charts, timelines, and even a replay button.
Here is what the tool actually does and what the researchers found. The visualizer allows teachers to upload a single log file, a whole folder of them, or a ZIP archive containing data from up to 87 students at once. Once uploaded, the system acts like a traffic controller, sorting through the chaos to show a "session dashboard" for each student. This dashboard features an interactive timeline that looks like a heartbeat monitor, showing exactly when a student typed, deleted, ran their code, or hit an error. It also includes a "code size graph" that tracks how the program grew (or shrank) over time, and a "replay" feature that lets teachers watch the student's coding process play out like a video, complete with pauses and bursts of activity.
The researchers suggest that this tool is a game-changer for spotting trouble. For instance, if a student's timeline shows a long period of silence followed by a sudden, massive jump in code size, it might suggest they copied and pasted a large chunk of code rather than writing it themselves. Similarly, if a student runs their program only at the very end of the session after hours of typing, it might indicate they were stuck and only tested their work when forced to submit. In a pilot test with 160 students and five teachers, the tool successfully flagged these suspicious patterns. The teachers found it incredibly useful for quickly identifying which students needed help and which cases required a closer look for academic integrity.
However, the paper is careful not to overhype the results. The authors explicitly state that the tool is designed to support teacher decision-making, not to replace it. It flags potential issues—like high amounts of pasted text or strange typing patterns—but it does not automatically declare a student a cheater. The tool suggests that these visual clues are powerful for starting a conversation, but the final judgment still belongs to the human teacher. The study also notes a limitation: the tool only works if students use a specific program called Thonny and submit their logs, meaning it can't analyze students who use different tools or forget to save their data.
In short, the researchers found that by using AI to quickly build a "vibe-coded" visualizer, they could turn invisible digital footprints into a clear, navigable story. This helps teachers move from guessing what happened to seeing the actual process, allowing them to offer help sooner and understand the "how" behind the final grade. While the tool isn't a magic wand that solves every classroom mystery, the study suggests it is a highly effective, low-barrier way to bring transparency to the coding process, making it easier to support students and maintain honesty in large classes.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.