View Proposal

---

Proposer

Ian Tan

Title

Low-Code Customisation of Open-Source Systems: Prompt-Driven Frontend Synthesis

Goal

To thoroughly document each stage of the prompt-driven engineering lifecycle, evaluating the reliability, architectural consistency, and limits of GenAI-guided software generation when anchored to an established open-source framework.

Description

This project explores an "almost zero-coding" paradigm to rapidly extend or modify robust, enterprise-grade open-source applications. Instead of manual software engineering, the student will act as a systems architect and AI orchestrator, studying the Frappe Framework alongside one of its specialised upstream applications (CRM, HR or Helpdesk). The core objective is to bypass traditional coding by leveraging Generative AI (GenAI) to synthesise fully functional frontend interfaces directly from visual wireframes and UI drawings (or hand drawn UI), using the system's underlying metadata as an operational context. This project requires the design, documentation, and evaluation of three distinct components: * The Process Layer (Activity Diagrams): Mapping out precise process flows, system interactions, and logical data states using formal activity diagrams or flowcharts to define how the new interface behaves. * The Interface Layer (Prompt-Driven Architecture): Translating physical or digital drawings of the frontend into production-ready UI code via GenAI. The student must establish an appropriate frontend architecture, construct a specialised "AI Context" containing the inner workings and API schemas of Frappe, and engineering iterative prompts to connect, populate, and test the synthesised frontend against the Frappe backend. * The Insights Layer (Visual Analytics): Creating a comprehensive, human-in-the-loop executive dashboard built entirely within Frappe Builder to analyse app-generated data traces, tracking operational performance or engagement metrics without traditional development overhead. (This may not be complete as you will not have enough data but the initial attempt to provide it alongside with the application).

Resources

* Kebaili, Z. K., et al. (2024). "An Empirical Study on Leveraging LLMs for Metamodels and Code Co-evolution." Journal of Object Technology, 23(3). * Babur, Ö., et al. (2026). "Selling Shovels in the LLM Gold Rush..." Journal of Object Technology, 25(2).

Background

Url

Difficulty Level

Moderate

Ethical Approval

None

Number Of Students

1

Supervisor

Ian Tan

Keywords

vibe coding, software engineering, front-end

Degrees

Bachelor of Science in Computing Science