quiz-designer (Quiz Content Design)
When to Use
- Need to design a "quiz" from scratch:dimensions / questions / scoring / result page / distribution any part.
- The user provides a theme or reference case (such as SBTI), and needs output Markdown / JSON / full copy specifications.
- Need to distinguish result forms and choose the right methodology (see the table below, including extensions such as "personality pattern matching").
Instructions
- Determine the type: first lock in one of the types in the table below by "result form", then expand dimensions and question count; do not mix scoring logic.
- Determine dimensions: 2–8 dimensions; define polarity, number of types (4/8/16, etc.), and whether easter eggs/hidden types are needed (about 10%–20% trigger probability).
- Write questions: question stems should be colloquial and situational; 2–4 options with discriminative power; control dirtiness; add validation questions when necessary to prevent random selection.
- Write scoring: explain dimension mapping, weights, normalization, ties, and fallback rules.
- Write the result page: label + short description + action hook; provide suggested share copy and topic tags.
- Write distribution: share triggers, pre-planted comments, remix/poster and other touchpoints.
Quiz Type Taxonomy
Based onresult form, divided into the types in the table below (including extensions); when designing, first determine the type, then choose the corresponding methodology:
| Type |
Result Form |
Typical Case |
Scoring Logic |
| Personality type type |
Discrete labels (such as INTJ / Creator) |
MBTI, sixteen types |
Elimination method: items correspond to types, eliminating others to lock in the result |
| Continuous spectrum type |
Continuous scores or bands (such as L/M/H per dimension) |
Emotional dependence, etc. |
Dimension stacking + normalization or banding, then mapping to intervals / patterns |
| Personality pattern matching type(typical extension) |
Discrete labels + multi-dimensional band vector compared with patterns in the library |
SBTI(25 regular + fallback + easter egg) |
Score per dimension → L/M/H → calculate distance and similarity with preset 15-position "standard patterns", rank and take the best match |
| Knowledge quiz type |
Accuracy/score |
Encyclopedia challenge |
Score for correct answers, cumulative or weighted sum |
| Numerical calculation type |
Specific numerical value (such as "buying a house takes X years") |
Financial freedom calculator |
Formula calculation, result determined by input variables |
Core Competency Points
Psychological Foundation (Brief)
- Three Major Gaps: Self-affirmation (who I am), social currency (comparable labels), certainty comfort (giving quantified conclusions). If any one is missing, shareability drops noticeably.
- Essence of selling: What users forward is the feeling of "being seen"; the result must simultaneously serve labels, resonance, and discussability.
- Edginess: Questions can be made life-oriented or even meme-like to boost participation; for sensitive questions, use safe options or downweight to avoid runaway controversy.
Type spectrum design
- Assessment dimensions: 2–8; set dimension polarity (positive/negative or two opposing ends).
- Number of types: 4/8/16, etc.; too many causes shareability decay; easter eggs/hidden types can be designed.
Question design
- Question stem: colloquial, situational, avoid preaching.
- Options: 2–4 items; discriminative; for sensitive questions, set safe options or downweight.
- Guidance: warm-up at the beginning, transition in the middle, and a light hint at the result at the end.
- Anti-random-clicking: two questions on the same dimension with different wording for consistency checking.
Scoring mechanism
- Dimension scoring: each question maps to 1–2 dimensions, weighted sum.
- Normalization: raw score for each dimension → 0–100 standard score (if needed).
- Classification: for type-based, use the highest dimension/elimination method; for spectrum-based, use score intervals.
- Tie: when tied, prioritize recommending a positive result (unless the product has special rules).
Result page and sharing
- Result page: title (type name + emotional hook), 2–3 sentence description, keyword tags, action hook.
- Sharing: share triggers, topic embedding, comment section operations, posters/nine-grid/topic tags.
Invocation convention (slash commands)
/quiz-designer <主题> [选项]
| Option |
Description |
--type |
Result type:personality / spectrum / knowledge / calc |
--dimensions |
Number of dimensions, default 4 |
--cases |
Reference case, e.g. SBTI |
--output |
Output format:markdown / json / full(including complete copy) |
Example:
/quiz-designer "职场性格测试"
/quiz-designer "恋爱人格测试" --type personality --dimensions 4
/quiz-designer "你在北京实现财务自由要多久" --type calc
/quiz-designer "创意能力测评" --output full
Default deliverables
- Type spectrum chart(dimensions and type definitions)
- Question list(question stems + options + scoring rules)
- Scoring algorithm explanation(dimension weights, classification logic)
- Result page copy template(title + description + sharing text)
- Sharing node planning(release timing, topic tags, interaction design)
Case: SBTI (personality pattern matching type)
SBTI testis a fun personality test circulating online (the name is similar in style to MBTI, butit is nota four-letter E/I, S/N, T/F, J/P voting type). The official implementation is:15 sub-dimensions, each dimension has 2 questions × 3 options (1/2/3 points) → aggregated into L / M / H three tiers, yielding a 15-position level sequence, then compared for similarity with the preset "standard patterns" of 25 regular personalitiesin the library; there are also fallback personalitiesandhidden easter egg personalities. When designing a test, it can be classified as "multi-dimensional tiering + distance between vector and standard answer."
Timeline and reference entry points (for traceability)
Fifteen-dimensional structure (five major blocks × 3 dimensions each)
| Model |
Dimension code |
Meaning |
| Self model S |
S1 / S2 / S3 |
Self-esteem and confidence, self-clarity, core values |
| Emotion model E |
E1 / E2 / E3 |
Attachment security, emotional investment, boundaries and dependence |
| Attitude model A |
A1 / A2 / A3 |
Worldview tendency, rules and flexibility, sense of life meaning |
| Action drive Ac |
Ac1 / Ac2 / Ac3 |
Motivation orientation, decision-making style, execution mode |
| Social model So |
So1 / So2 / So3 |
Social initiative, interpersonal boundaries, expression and authenticity |
Scoring and classification (consistent with the official version)
- Answering: 2 questions per dimension, each question choosing A/B/C corresponds to 1 / 2 / 3 points; the raw score for that dimension ∈ [2, 6].
- Tiering：2–3 → L，4 → M，5–6 → H. The 15 dimensions in order yield a string of 15 letters, such as
H H L - H M H - …(spaces may be added or it may be written continuously when displayed).
- Pattern comparison: treat L/M/H as 1 / 2 / 3, and compare with the user vector and each personality's 15-position standard patternposition by position, taking the absolute value of the difference and summing them as the "total distance". The maximum theoretical distance is 15×2=30。
- Similarity：
相似度 = (1 − 总距离/30) × 100%。
- Selection: calculate the distance one by one against the 25 regular personalities,smaller distance takes priority; if distances are equal, more exact matchestake priority; if still equal, then higher similaritytakes priority. The first place is the main result personality.
- Fallback: if the first place's similarity isbelow 60%, the result is HHHH (Silly Joyful One)。
- Easter egg: if the specified option is triggered in the officially defined "special question" path,skip the above matching, and directly make it DRUNK (Drunkard)(see
.claude/commands/sbti.md）。
Overview of regular personalities (code + 15-position standard pattern)
Display names and patterns for easy comparison during implementation; the complete opening remarks and long copy are subject to the official question bank/page.
| Code |
Chinese label |
Standard pattern (S-E-A-Ac-So five groups, written continuously without separators) |
| CTRL |
Controller |
HHH-HMH-MHH-HHH-MHM |
| ATM-er |
Money Giver |
HHH-HHM-HHH-HMH-MHL |
| Dior-s |
Loser |
MHM-MMH-MHM-HMH-LHL |
| BOSS |
Leader |
HHH-HMH-MMH-HHH-LHL |
| THAN-K |
Grateful One |
MHM-HMM-HHM-MMH-MHL |
| OH-NO |
Oh No Person |
HHL-LMH-LHH-HHM-LHL |
| GOGO |
Walker |
HHM-HMH-MMH-HHH-MHM |
| SEXY |
Bombshell |
HMH-HHL-HMM-HMM-HLH |
| LOVE-R |
Romantic |
MLH-LHL-HLH-MLM-MLH |
| MUM |
Mom |
MMH-MHL-HMM-LMM-HLL |
| FAKE |
Fake Person |
HLM-MML-MLM-MLM-HLH |
| OJBK |
Whatever Person |
MMH-MMM-HML-LMM-MML |
| MALO |
Monkey |
MLH-MHM-MLH-MLH-LMH |
| JOKE-R |
Clown |
LLH-LHL-LML-LLL-MLM |
| WOC! |
WTF Person |
HHL-HMH-MMH-HHM-LHH |
| THIN-K |
Thinker |
HHL-HMH-MLH-MHM-LHH |
| SHIT |
Cynic |
HHL-HLH-LMM-HHM-LHH |
| ZZZZ |
Playing Dead Person |
MHL-MLH-LML-MML-LHM |
| POOR |
Impoverished One |
HHL-MLH-LMH-HHH-LHL |
| MONK |
Monk |
HHL-LLH-LLM-MML-LHM |
| IMSB |
Fool |
LLM-LMM-LLL-LLL-MLM |
| SOLO |
Orphan |
LML-LLH-LHL-LML-LHM |
| FUCK |
Grass Person |
MLL-LHL-LLM-MLL-HLH |
| DEAD |
The Dead |
LLL-LLM-LML-LLL-LHM |
| IMFW |
Loser |
LLH-LHL-LML-LLL-MLL |
Special Personality
| Code |
Name |
Condition |
| HHHH |
Happy Fool (fallback) |
The highest similarity with all regular personalities is still below 60% |
| DRUNK |
Drunkard (hidden) |
In the special question chain, select official preset options such as "Pour baijiu into a thermos and drink it as plain water" |
State machine (executed within the conversation)
INIT → QUESTIONS → RESULT → DONE
INIT: Briefly explain the SBTI rules and number of questions (30 regular questions + official special questions),q=1。QUESTIONS: Display the questions in official order with A/B/C options, recording 1/2/3 points; when encountering special question branches, handle according to sbti.md .RESULT: Generate the scored vector → ranking / fallback / easter egg determination → output the corresponding personality'sopening statement + description(can be excerpted from official copy).DONE: Guide sharing and retesting; reference links may be attached.
Interaction and Input
- User initiates:
/sbti or equivalent entry point.
- Accepts:
A / B / C、(A), Chinese "choose A", etc., case-insensitive (consistent with the official three options).
Example result page structure (not MBTI four letters)
## 🎯 你的 SBTI 人格：THIN-K（思考者）
**开场白**：已深度思考100s。
**相似度**：93%（示例）
### 十五维档案（L / M / H）
| 模型 | 维度 | 你的档位 |
|------|------|----------|
| S | S1 S2 S3 | … |
| … | … | … |
### 正文节选
> （粘贴该人格官方描述段落）
---
✨ 测试完成！想再测一次请说「重测」
Sharing hooks (examples)
- Share text:
我测出来是 THIN-K，你也来试试？
- Topic:
#SBTI #测一测
- Pre-seeded comments:
晒出你的人格代号
Question bank structure specification (aligned with SBTI)
- questions per dimension; 2 options per question, with point values 3 ; for question text and option copy, see 1、2、3full table.
.claude/commands/sbti.md full table.
- When designing similar quizzes: first lock down the dimension definitions, then ensure the number of questions per dimension is symmetrical, and finally maintain a "standard mode" vector for each result and clearly write out the distance and fallback rules.
Methodology references
- Barnum effect / Forer effect; MBTI; Big Five personality (dimension reference); 5W of communication (communication nodes).
