Agent-Skills.md

Agent Skills: Conversation Quiz

Generates multiple choice quiz questions based on current conversation context. Use when testing understanding, reviewing what was discussed, or wanting a knowledge check on the session.

UncategorizedID: antoniocascais/claude-code-toolkit/quiz

Repository

antoniocascais/claude-code-toolkit
antoniocascaisLicense: AGPL-3.0
5

Install this agent skill to your local

pnpm dlx add-skill https://github.com/antoniocascais/claude-code-toolkit/tree/HEAD/skills/quiz

Skill Files

Browse the full folder contents for quiz.

Download Skill

Loading file tree…

skills/quiz/SKILL.md

Skill Metadata

Name
quiz
Description
Generates multiple choice quiz questions based on current conversation context. Use when testing understanding, reviewing what was discussed, or wanting a knowledge check on the session.

Conversation Quiz

Generate multiple choice questions testing understanding of the current conversation.

Instructions

  1. Analyze the conversation context for quizzable content:

    • Technical concepts discussed
    • Decisions made
    • Code patterns or implementations
    • Key facts or configurations

  2. Generate 3-10 questions with 4 options each (default 5, or as many as the user/arguments request)

  3. Question Quality Guidelines

    • Questions should test understanding, not memorization of exact wording
    • Include plausible distractors (wrong answers that could seem right)
    • Vary difficulty: mix straightforward recall with deeper comprehension
    • Descriptions should explain WHY the option is correct/incorrect (shown only after answering)
    • If conversation lacks substance for 3 questions, generate what's reasonable and note the limitation

  4. Write questions to file and launch external quiz runner:

    • Generate a unix timestamp: date +%s via Bash
    • Write questions JSON to /tmp/quiz_questions_$TIMESTAMP.json using the format below
    • Tell the user to run in a separate terminal: python3 ~/.claude/skills/quiz/quiz.py /tmp/quiz_questions_$TIMESTAMP.json
    • The quiz runner writes results to /tmp/quiz_results_$TIMESTAMP.json (same timestamp, auto-derived from questions filename)
    • Wait for user to report back that they finished
    • Read results from /tmp/quiz_results_$TIMESTAMP.json
    • Provide feedback: celebrate correct answers, explain wrong ones with the description from the correct option

Questions JSON Format

[
  {
    "question": "What network mode allows containers to share the host's network namespace?",
    "options": [
      {"label": "bridge", "correct": false, "description": "Bridge creates an isolated network — containers get their own namespace"},
      {"label": "host", "correct": true, "description": "Host mode removes network isolation — container shares the host's network stack"},
      {"label": "overlay", "correct": false, "description": "Overlay enables multi-host networking, but still uses separate namespaces"},
      {"label": "macvlan", "correct": false, "description": "Macvlan assigns a MAC address to the container — separate namespace with direct network access"}
    ]
  }
]

Important

  • Each question MUST have exactly ONE option with "correct": true
  • The description field is NOT shown during the quiz — only used for post-quiz feedback
  • Options are shuffled automatically by the quiz runner — no need to randomize in the JSON
  • The quiz runner handles: display, input, timing, scoring, and writes results to file