Skip to content
Back to skills

Quiz

ASecurity

Pure retrieval session, no teaching - due reviews and Error Log redos, weighted to older and previously failed notes, with topic labels hidden. Use when the learner asks to be quizzed or to review what is due.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 18, 2026
ai-agentsgonode

Security analysis

A100/100

Scanned November 18, 2026

npx -y skills add jlu1023/learning-stack --skill quiz --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Quiz?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Quiz
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/jlu1023-quiz/badge)](https://www.skillsdirectory.com/skills/jlu1023-quiz)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: quiz
description: Pure retrieval session, no teaching - due reviews and Error Log redos, weighted to older and previously failed notes, with topic labels hidden. Use when the learner asks to be quizzed or to review what is due.
argument-hint: "[course | mixed] [n]"
disable-model-invocation: true
---

Run a retrieval session for `$ARGUMENTS` (default: `mixed 8`).

1. **Load the rules.** Load the `method` skill (Primers and the week ahead, Questions, Marking, Spacing and interleaving, Logging, Tone) and read `Courses/_learning/Learner Profile.md` and `Courses/_learning/Teaching Style.md` (for course overrides and question format). `VL` means the engine command in the method skill.

2. **Week ahead.** If this is the first `teach` or `quiz` run of a new week, give the week-ahead overview first (method, Week ahead).

3. **Draw the items.** Run `VL due <course|mixed>`. It returns due nodes, already ordered by overdue days plus logged errors, and due Error Log redos.
   - Take `n` items (default 8). Include every due redo first, then the due nodes in order, which weights them towards older and previously failed nodes.
   - For `mixed`, interleave courses: never two items from the same module in a row when avoidable.
   - If fewer than `n` are due, say how many and run only those. Do not pad with unseen material; that is what `teach` is for.
   - `due` never lists proofs: they are practised on the proof track (`teach <course> proofs`). If `VL proofs mixed` shows proofs due, say how many at the close.

4. **Ask, one item at a time,** with **no course, module or topic labels**: not in the question, options or preamble.
   - Build each question from the node's note (for a card, its prompt and answer, checked against its source), dressed in a cue class per the method (Varying the cue): `learning` gets `lecture` or `notation`; from `secure` up, a class other than the node's last cue (from the session files), the recall dresses before `twin` before `combined`, with `reverse` and premise-break questions from the node's chain; a node due to become `mastered` gets a teach-back (method). Pitch it at the item's `held_as` line when there is one. Never ask for a proof to be written: for a theorem, at most its proof's key idea or chain in one line.
   - Adjust as you go: a clean pass moves that node one class further along next time, a pass that needed a hint holds, a miss moves it back. Multiple-choice distractors each name a wrong picture (method, Questions).
   - Mark per the method (Marking): the recall stage for `unseen` and `learning` nodes; from `secure` up, sort each miss and retest wrong-tool and misconception misses at the end.
   - A redo reworks the logged mistake with fresh numbers, and it is labelled "(generated)" in the log.
   - Multiple choice goes through `AskUserQuestion`. Precise statements are free response. Apply any course override in Teaching Style (for example, statements and methods only for a methods course).
   - For multi-step problems, the learner restates the problem from the stem first.
   - Confidence comes with the answer (method, Questions), then the verdict, then name the node. Flag overconfidence.
   - No teaching: on a wrong answer, point at the first broken line, give the correct statement in one or two lines, and move on. Offer `teach` for that node at the end, and name its unit's primer if it has one (`VL graph <COURSE>` gives the node's `unit`, `VL unit <COURSE> <unit>` its `primer`), rather than re-explaining.

5. **Record.**
   - Run `VL record "<node>" --result pass|fail` after each item, and note its cue class beside the question in the session file.
   - At the close, append a Learner Model line for each node whose answers showed how the learner holds it (method, Logging). Silent: do not mention it in chat.
   - Append each mistake to `Courses/_learning/Error Log.md` in the method's format.
   - When a redo is answered correctly, append ` [redone:: YYYY-MM-DD]` to its Error Log line.

6. **Close.**
   - Write `Courses/_learning/sessions/YYYY-MM-DD <course|mixed> quiz.md` (`type: session`, the method's format, no plan diagram).
   - Run `VL roadmap all` (or the one course) to refresh the colours.
   - Append a `quiz | <course|mixed>` entry to `_meta/log.md`.
   - Report in chat: the score, overconfident misses, notes that dropped a level, what is due next, and the number of proofs due (`VL proofs mixed`).

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in ai-agents →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.