SAQ

What SAQ practises

SAQ is a dedicated specialist in the KoiSwarm agent selector, built for concise written-recall practice — the kind that tests whether you can produce an answer, not just recognise one among options. Give it a topic or source material and it returns short-answer questions with a marking guide for each.

How a set is built

1. Read the brief. SAQ takes the source or topic, question count, expected answer length, and mark allocation you specify.
2. Write questions that require recall. Each question is phrased so it cannot be answered by pattern-matching a multiple-choice stem — you have to produce the answer yourself.
3. Build the marking guide. Every question ships with an explicit, source-grounded guide showing what earns each mark.
4. Mark your response. After you answer, your response is compared against the marking guide and scored point by point.

Write a useful request

Specify the source or topic, number of questions, expected answer length, and how feedback should be returned. For example:

“From this renal lecture, give me four five-mark short-answer questions. Show one question at a time, then compare my response with an explicit marking guide.”

Continue your practice

A saved SAQ session can be reopened from its history or the conversation that created it. It keeps the questions, submitted answers and feedback together, so returning to an unfinished set does not require a new brief.

When you open an SAQ task from Goals, KoiSwarm prepares the questions and source-grounded marking material before opening the session. The task’s scope and linked teaching files carry through to the practice, and the marked work can contribute evidence to that Goal.

Review the result

Short-answer practice should test recall without recognition cues. Check that the requested marking guide is supported by the source and treat generated grading as formative feedback rather than an official academic result.