Agentic AI Vibe Coding Workshop at Crescent Girls' School

Building Real Projects with Agentic AI

Crescent Girls’ School engaged Mind Theory to run an Agentic AI Vibe Coding Workshop for their Computer CCA, with Sec 1-3 students taking part across 4 x 2hr sessions. Every session paired concepts with hands on practice, so students finished the programme having built 4 working projects of their own: an online quiz, an HTML game, a note taking app, and their own website. (Schools can read more about the full programme on our Agentic AI Coding Workshop page.)

Lesson 1: What is Agentic Coding

The first session opened with a look at what AI is and what it can already do, before introducing agentic AI: AI that can plan, take actions, and complete multi step tasks on its own rather than only answering questions. Students saw real examples of the benefits, then compared the leading tools in a comparative overview of Google Antigravity, OpenAI Codex, Claude Code, GitHub Copilot in VS Code, and MTCode.

We also covered the difference between an IDE (Integrated Development Environment) and a CLI (Command Line Interface), and how AI assisted development differs from traditional line by line coding. Students then went hands on with the MTCode terminal CLI, following a live project demo to build their first project: an online quiz.

Lesson 2: Prompting Frameworks

Session 2 focused on the skill that matters most in vibe coding: writing clear, structured prompts. Students learned Google’s 5 Step Prompt Framework and worked through example prompts showing how specific, structured instructions produce far better results than vague ones.

They were also introduced to Markdown files and Skills, the building blocks agentic tools use to store instructions and reusable behaviours. Continuing in the MTCode CLI, students put the framework to work on their second project: building their own HTML game.

Lesson 3: Moving into an IDE

Session 3 moved students from the CLI into a full IDE. They signed up for their own free GitHub accounts, explored the GitHub Copilot in VS Code interface, and learned when to use its Plan, Ask, and Build modes.

Their third project was a note taking app. Students described the app they wanted in natural language, watched Copilot build it, then refined the output through iterative instructions and debugging. Testing and improving their own app taught them the core loop of AI assisted development: describe, build, test, refine.

Lesson 4: Shipping Their Own Website

The final session continued in GitHub Copilot VS Code, with students learning to modify their app’s interface, fonts, colours, logic, and features through prompting alone. Then came the highlight of the programme: each student designed and built a website of her own, on a topic she chose herself.

We closed with a discussion on ethical AI usage, weighing the pros and cons of building with AI, followed by a Q&A and a glossary recap of the terms covered across the 4 sessions.

What the Students Built in 4 Sessions

Here are some selected student projects, live and working exactly as the students shipped them. Click a card to load it right here, or use the full screen link under each one.

Personal Sites

▶ View Site

Annabelle’s Personal Site – by Annabelle  |  Open full screen  |  Close

▶ View Site

Nincompoop’s Pixel Palace – by Claudia  |  Open full screen  |  Close

▶ View Site

Welcome to My World – by Zoey  |  Open full screen  |  Close

Games

▶ Play

Pizza Speed Game – by Amet  |  Open full screen  |  Close

▶ Play

Block Shooter – by Balaji  |  Open full screen  |  Close

▶ Play

Flappy Bird – by Charmaine  |  Open full screen  |  Close

▶ Play

Cozy Penguin Game – by Charmaine  |  Open full screen  |  Close

▶ Play

NEON PONG – by Delfyne  |  Open full screen  |  Close

▶ Play

Chess vs AI – by Emily  |  Open full screen  |  Close

Tools

▶ Open Tool

SG Job Finder – by Celine  |  Open full screen  |  Close

▶ Open Tool

Time Tracker v1.0 – by Zoey  |  Open full screen  |  Close

Browse every app and game our students have built on our Student Showcase.

Reviews & Feedback

“I enjoyed being able to create a website of my own with my own ideas. The actual coding itself wasn’t that hard because it’s just prompting. This program is very informative and allows students to know how they can easily prompt AI, in detail, to code something for them.” – Sarah

“I enjoyed being able to brainstorm and create our own projects and see the webpage being created. The content was easy to understand because using AI to vibe code is quite straightforward, but I learned how to prompt AI in a more specific manner. The trainer was quite helpful whenever we needed assistance.” – Delfyne

“I enjoyed using MTCode to make games and websites. The content was definitely easy to understand as I carried out the project easily. The trainer was patient and kind. I would recommend it to any of my friends.” – Zoey

Summary

This Agentic AI Vibe Coding Workshop gave the Computer CCA students at Crescent Girls’ School a genuine head start in the way software is now being built. Across 4 sessions, every student went from her first prompt to a finished website, learning to direct AI tools with clear, structured instructions rather than writing every line of code by hand. For a deeper look at why vibe coding belongs in every school’s enrichment calendar, read our post on Vibe Coding as an enrichment skill.

If you’d like us to run this programme for your school, enquire now via Email or Whatsapp.