All products

Design a classroom exam with AI

Build a paper around the learning outcomes, then prepare a separate key and marking guide. Review individual questions afterwards using anonymised response data.

To design a classroom exam with AI, decide what you need to learn from the results. A student might recognise a definition without being able to apply it to a new case. The exam-construction skill helps you make room for both in a blueprint before the first question is written.

A lecturer can then inspect each question alongside its intended purpose. For a multiple-choice item, the guide asks why a student with partial understanding might choose each distractor. For a written response, it sets out which parts of the reasoning earn marks. The student paper and marking materials are separate outputs.

What it does

  • Check whether the time and marks fit the course priorities.
  • Draft a case that requires students to apply an idea to unfamiliar evidence.
  • Revise a question that admits two plausible correct answers.
  • Prepare comparable A and B papers from the same blueprint.
  • Review omissions and option frequencies after marking the exam.

Install and use

  1. Install the skill and give the agent the syllabus, learning outcomes and time available. Python is needed only for the optional local checks. Specify the output language, as the instructions default to Chinese.
  2. Ask, “Plan a 40-minute paper. Show me which questions assess application before drafting the marking guide.”
  3. Compare the paper with the key and try scoring one sample answer. Revisit any criterion that allows inconsistent marks.

FAQ

Source

Made by psychelzh. Original skill: exam-construction, published on GitHub under the CC-BY-4.0 licence. The files are unchanged.

View on GitHub