Skip to content

Project Addendum

This document clarifies the project requirements for our section and sets expectations for AI-assisted development. You are expected to use GitHub Copilot (with Education Benefits) throughout the project. Effective use of AI requires practice.

Learning Objectives

By the end of this project, you will be able to:

  • Write effective prompts to generate, debug, and refactor code.
  • Evaluate and modify AI-generated HTML, CSS, and JavaScript.
  • Coordinate AI-assisted development across a team using version control.
  • Reflect on your development process through documented chat logs.

Code Ownership

Each team member must hand-write at least 30% of the code they are responsible for. Your code need not be written from scratch; we expect you to review and modify AI-generated code. However, the project cannot be 100% “vibe coded.”

More importantly: every team member must be able to explain and modify any part of the codebase during a live code review. If you accepted AI-generated code that you don’t understand, go back and learn the code—or rewrite that code yourself.

How this is assessed

The 30% target is evaluated per person, not per team. Your instructor assesses this holistically from chat logs, commit history, and your ability to explain code during the presentation. Understanding matters more than authorship: if you can’t walk through how a section works and make changes on the spot, that’s a problem regardless of who (or what) wrote the code.

Coordinating AI Use

When multiple people use Copilot on the same project, you can easily end up with conflicting code, duplicated components, or inconsistent styles.

  • Divide by feature, not by file. Assign each team member responsibility for specific claims or features, not just “the CSS” or “the JavaScript.”
  • Agree on conventions early. Decide on naming conventions, file organization, and shared patterns (e.g., how you structure event listeners, how you access localStorage). Share these with Copilot as context.
  • Pull before you prompt. Always git pull before starting a work session so that you and Copilot see the latest code.
  • Review each other’s AI-generated code. During your weekly planning, briefly walk through what each person merged. This catches inconsistencies early and helps everyone learn.

Weekly Workflow

We recommend you follow this cycle as a team:

  1. Plan – Decide which claims to work on this week. Divide tasks among team members so that work doesn’t overlap or conflict.
  2. Prompt – Give Copilot specific, well-scoped requests. Reference your existing code and project context. Avoid vague prompts like “build me a dashboard.”
  3. Review – Read and understand every line of code you accept. Test in the browser. If something doesn’t make sense, ask Copilot for an explanation.
  4. Commit – Make small, frequent commits with descriptive messages. Each student should have multiple commits per week.
  5. Export – Export your chat logs and commit them alongside your code changes.

Aim for two non-Basic claims per week

You need at least 10 non-Basic claims for an A grade, and you have 5 weeks. Don’t back-load everything; claims cannot be made during finals week. And you can’t make more than 3 claims in a single week.

Commit Chat Logs

Export and commit your Copilot chat sessions to the chat/ folder in your repository at the end of each work session. A copy of export_chats.py from vscode-chat-export has been added to your repo.

Your chat logs are part of your grade. Your instructor will review them to assess your development process, not just your final product. Thoughtful, iterative conversations with Copilot demonstrate learning. Shallow or copy-paste-and-move-on logs do not.

What good chat logs look like

  • You describe what you’re trying to build before asking for code.
  • You ask follow-up questions when you don’t understand the output.
  • You request changes or alternatives rather than accepting the first result.
  • You use Copilot to debug errors and explain unfamiliar code.
  • See also Best practices for using AI in VS Code in the documentation.

Developing Your Skills

The goal of this addendum is not just to build an app. You are learning how to build web applications with AI assistance. Each week, push yourself to use Copilot in a new way. Here are some examples:

Week Skill to Practice
10 Use Copilot to scaffold components from your design sketch. Practice writing specific prompts.
11 Use Copilot to integrate your third-party API. Ask Copilot to explain the fetch/response cycle.
12 Use Copilot to debug issues and refactor code. Ask “why isn’t this working?” instead of starting over.
13 Use Copilot to improve accessibility and responsive design. Ask Copilot to audit your HTML.
14 Use Copilot to polish: transitions, edge cases, data export/import. Prepare for your presentation.

Project Pacing Guide

You are expected to make incremental progress. With five weeks, you should complete 20% of the project each week. The following table outlines one way to get an A on the project.

Date Target Progress (cumulative)
Apr 07 20% Project started and many Basic claims complete.
Apr 14 40% All Basic claims complete. 2 Intermediate claims.
Apr 21 60% 4 Intermediate claims. API integration and persistence working.
Apr 28 80% 6 Intermediate claims + 1 Stretch. Responsive and accessible.
May 05 100% 8 Intermediate + 2 Stretch claims. Presentation rehearsed.