Student Grouping Generator

You are a classroom teacher who has learned that "just pick randomly" is harder than it sounds once fairness actually matters, a student who's been called on five times while another hasn't been called on once, or a group that ends up lopsided in skill level when the goal was a mix. This tool handles three different grouping jobs, and each one needs a different approach.

Here is my class roster: [ROSTER]. If there's a specific goal shaping how students should be grouped, mixed ability levels, students who work well together, students who need to be kept apart, note it here: [GROUPING_GOAL?].

Set [MODE:select:pick one student at random,form groups,create a call-order for the whole class] to choose the job.

1. If pick one student at random, choose one name from [ROSTER] and say it plainly. If I've told you who has already been picked recently in [GROUPING_GOAL?] or elsewhere in this conversation, weight the pick away from students who've gone recently so the same few names don't dominate over a week.

2. If form groups, build groups from [ROSTER] sized to fit [GROUPING_GOAL?] if I specified a group size or number of groups, otherwise use groups of four as a default. If [GROUPING_GOAL?] calls for mixed ability or mixed personality types and I've given you information to work from, distribute students so no group is accidentally stacked. If [GROUPING_GOAL?] names students who need to be kept apart, honor that constraint over every other consideration, even if it makes the groups less evenly sized.

3. If create a call-order for the whole class, build a full random order covering every name in [ROSTER] exactly once, so cold-calling for a discussion or a presentation schedule hits everyone before anyone repeats. State the order plainly, numbered.

4. Whichever mode I chose, if [ROSTER] looks incomplete or has an odd structure, ask me to confirm rather than guessing at missing names, since a grouping tool that quietly drops a student is worse than one that flags the gap.

Close by naming any fairness trade-off you had to make, such as prioritizing a keep-apart constraint over a perfectly even group size, so I know the reasoning behind the result instead of just the result itself.