FlowCheck: Helping End-Users Specify and Verify Intent in Vibe-Coded Web Apps
Source
arxiv.org
Author
Reya Vir, Lydia Chilton, Zhuo Zhang, Eugene Wu
Date
Context
FlowCheck lets someone specify how information should move through an AI-generated web app by selecting interface elements. A discount button, for example, should change both the displayed total and the saved total. The system converts those expectations into CodeQL checks that inspect the code without running it. In four small JavaScript applications, it detected all 30 deliberately introduced violations with no false positives. The implementation cannot fully check interfaces assembled dynamically by frameworks such as React, so the result establishes a useful way to verify specified information flows within a limited application structure. System, evaluation and limitations.
Terms in this piece · Glossary
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
articleVibeCheck: Assessing the Quality of LLM-Generated Unit Tests: A Multi-agent Empirical Study across Heterogeneous RepositoriesAnika Tabassum, Mushahid Intesum, Md. Fahim Arefin, Tarannum Shaila Zaman
articleSoundness Checking of Taint Flow ModelsSamarth Kishor, Victor Nicolet, Joey Dodds
articleVibe Coding: Practice, Performance, Productivity, and Risk -A State-of-the-Art ReviewDominik L. Michels, Mutaz Abu Ghazaleh, Francois Lazzari, Nabil Kassem, Jonathan Klein