AI Agent Teaching Applications Workshop (2): From Teaching Materials to a Self-Practice Tool

Smart Rainforest Industry Talent Development Program | This page is the handout for Workshop (2), compiled from planning documents; the official announcement for each cohort governs actual dates.

Course Information

Both sessions cover identical content; attend whichever session you registered for. Actual registration status and capacity follow the official announcement.

Agenda:

1. Series Positioning and Context

This is the second of three workshops in the “AI Agent Teaching Applications Workshop” series, designed for university faculty without an engineering background. We begin with a short review: an ordinary chatbot mainly responds in conversation, while an AI Agent can use files, tools, and a sequence of steps within a defined scope to complete a task. This session applies that collaborative approach to two real workflows: building a question bank and building a self-practice web page.

This session builds on the AI collaboration method from Workshop (1): state the goal, source data, constraints, and completion criteria clearly; check the result; then turn the verified workflow into a reusable skill. Participants take away not only one-time outputs, but also a repeatable way of working.

2. Hands-on Flow

Stage 1 begins with public examination materials. Participants use an AI Agent to help download the required examination PDF and answer materials from the official site. They then split the PDF into one image per question and map each image to its correct answer, producing a question bank that can be stored and used locally.

Immediately after Stage 1, participants package and test the download, splitting, organization, and verification steps as an independent “question-bank pipeline skill.” Later, when building a question bank for another subject or year, they can apply that skill to execute the same specification precisely, without rediscovering the approach or unnecessarily spending time and tokens.

Stage 2 starts from the completed question bank and asks how to create a self-practice environment for students. Participants use a static-web-page approach and embed the question images in the HTML. Each test or practice unit is a self-contained HTML file that does not depend on an external image library or other files. This makes testing, sharing, and publishing much simpler.

Immediately after Stage 2, participants package and test the practice-unit generation workflow as a second independent “practice-unit generator skill.” Once a new question bank is prepared, that skill can directly create further practice units, keeping development repeatable, verifiable, and efficient.

3. Who Should Attend

• Faculty who attended Workshop (1), or who can briefly describe a task using “goal, context, constraints, and completion criteria.”
• Faculty who want to turn public examination questions, past papers, or existing question materials into a manageable local question bank.
• Faculty who want students to use a web page to practice repeatedly at their own pace after class.
• Faculty who prefer a small, complete, directly publishable static web solution before attempting a complex system.

4. Learning Outcomes

After completing this session, you will be able to:

1. Explain the difference between an AI Agent and an ordinary chatbot, and confirm that ChatGPT Codex, VS Code, and optional GitHub education resources are ready to use.
2. Use an AI Agent to download examination questions and answer materials, split a PDF into per-question images, and build a local question bank with answer mappings.
3. Immediately after Stage 1, package and test the verified question-bank workflow as an independently reusable skill.
4. Design a student self-practice environment from a prepared question bank and create standalone static HTML practice units with embedded images.
5. Immediately after Stage 2, package and test the practice-unit generation workflow as a second independent skill for direct reuse with other question banks.

5. Two Reusable Workflows

Workflow 1: build a local question bank from a PDF

• Specify the target year, subject, and public source.
• Download the examination PDF and answer materials.
• Split the examination into per-question images, with clear names and storage.
• Map the correct answer to each question and produce a local question bank for the practice page.
• Immediately package and test this workflow as an independent question-bank pipeline skill.

Workflow 2: build standalone practice units from a question bank

• Plan the student practice flow using the question-bank data.
• Create one standalone static HTML file for each unit.
• Embed the unit’s required question images in its HTML, without relying on an external image folder.
• Open, test, and share it locally; publish it to a static website when needed.
• Immediately package and test this workflow as an independent practice-unit generator skill.

6. Preparation Checklist

• Bring a laptop and confirm you can use a web browser.
• Confirm that ChatGPT is installed and signed in, with Codex available.
• Install VS Code. If useful, apply for a GitHub account and GitHub Education to use an eligible GitHub Copilot plan.
• This workshop uses public examination questions and answer materials as examples; do not upload student personal data, grades, or other sensitive information.

To review the environment and setup basics beforehand:

Each core stage has a step-by-step hands-on tutorial that can be used as self-paced reference during or after the workshop:

7. Publishing and Reuse

This session’s output is intended as an after-class self-practice environment for students, not as formal grading or a high-stakes test. Each practice unit is an independent HTML file with its question images embedded, so there is no separate image library or complex website dependency to manage. It can be given directly to students, placed in a shared folder, or published on a static website.

• The key to reuse is not simply copying an output once, but keeping the verified skill so new data can be regenerated to the same specification.
• After each download, split, and answer-mapping run, spot-check question numbers, images, and answers for correctness.
• After each HTML page is generated, open it locally and test the question display and answer flow before sharing it with students.
• If student accounts, score collection, personal-data handling, or formal assessment are needed later, plan data governance, security, accessibility, operations, and institutional procedures separately.

8. Expected Takeaways

• A workflow and skill for building a local question bank from public examination PDFs and answer materials.
• A workflow and skill for building student practice units from a question bank.
• One or more standalone static HTML practice units with embedded question images.
• A set of checks for verifying the question bank and web pages.
• An AI Agent collaboration method that can be extended to other years, subjects, materials, or teaching tools.

9. Key Takeaways

The point today is not only to complete one question bank or one web page. Each completed stage is immediately turned into its own skill that an AI Agent can execute precisely and repeatedly. When the source material or question bank changes, participants do not have to explore from scratch; they can rebuild the output against clear inputs, outputs, and acceptance criteria.