Skip to content

Prompt Programming platform with bug injection pipeline

Design · All designs
Evidence · 2 claims (2 for) · 2 studies (1 causal, 1 associational), q3 · 0 of 2 report an effect size · 2 claims rest on one study

Description

A publicly available web platform supporting prompt-based programming through dialogue with a GenAI assistant (GPT-4o-mini), hidden-test execution, and direct code editing. The bug injection pipeline acts as middleware that intercepts correct generated code and replaces it with validated, runnable near-miss variants containing subtle faults. Students receive a problem specification, craft a natural-language prompt, and can iterate by reprompting or editing code directly. The platform was deployed in a CS1 course at the University of Auckland with five programming tasks covering arrays, matrices, and binary operations.

Design Implications

Context

Requirements

  • A GenAI code-generation model and hidden unit tests to validate that injected bug variants compile and fail

Constraints

  • Bug injection is limited to a set of candidate variants, permitting some slip-through cases where original correct code is served unchanged

Target Learners

  • CS1 introductory programming students

Learning Goals

  • code review
  • debugging
  • prompt specification
  • verification of AI-generated code

Claims

Examples

-

Key Sources

  • Victor-Alexandru Pădurean, Kaitlin Riegel, Alkis Gotovos, Jyotika Mahapatra, Ahana Ghosh, Paul Denny, Juho Leinonen, James Prather, and Adish Singla. 2026. When AI Is Wrong on Purpose: How Students Respond to Buggy GenAI Code. Proceedings of the ACM Conference on International Computing Education Research Vol. 1 (ICER 2026 Vol. 1). https://doi.org/10.1145/3765964.3811667
Page metadata
Field Value
type design
id prompt-programming-platform-bug-injection-pipeline
title Prompt Programming platform with bug injection pipeline
description A publicly available web platform supporting prompt-based programming through dialogue with a GenAI assistant (GPT-4o-mini), hidden-test execution, and direct code editing.
status review
generated by process:settle-candidates on 2026-10-10
trust unverified — no one has confirmed this page's content
sources 1 — victor-alexandru-pădurean-kaitlin-riegel-alkis

Frontmatter is the page's machine-readable record — see the Schema & Guide.