# Questionnaire v1

## Usage

- Run this questionnaire in the user's language. Translate the prompts, but keep the meaning stable.
- Start with intake. Skip questions the user already answered clearly.
- Ask 8 to 12 questions per round. Use only the rounds you still need.
- Accept option letters, short prose, or mixed answers.
- After each round, summarize the strongest patterns before continuing.
- If the user is getting tired, shorten the round and preserve only the highest-signal questions.

## Intake

Use these questions to ground the session before the five core dimensions.

1. What is the main failure mode of your current AI workflow?
   - A. It answers the wrong problem
   - B. It gives information without judgment
   - C. It uses the wrong format or tone
   - D. It does not turn ideas into a usable deliverable
2. What do you use AI for most often?
   - A. Research or synthesis
   - B. Product, strategy, or decision support
   - C. Writing or rewriting
   - D. Technical implementation or debugging
3. What do you want at the end of this skill?
   - A. One reusable master profile
   - B. Split files for different agents
   - C. A shorter summary plus operating rules
   - D. I am not sure yet
4. How structured should this session feel?
   - A. Mostly multiple choice
   - B. Mixed choice and free text
   - C. Mostly conversational
   - D. Fastest possible
5. What language should the final profile use?
   - A. Chinese
   - B. English
   - C. Bilingual
   - D. User-language body with English headings
6. What should a good AI collaborator do first?
   - A. Restate the real problem
   - B. Give a conclusion fast
   - C. Offer a framework
   - D. Ask clarifying questions

## Round 1: Cognitive Style

1. When you ask a complex question, what should come first?
   - A. Conclusion first
   - B. Overall framework first
   - C. Problem decomposition first
   - D. Background first
2. Which explanation style works best for you?
   - A. Mechanism chain
   - B. Framework and categories
   - C. Concrete example
   - D. Intuitive summary
3. How do you react to an answer that is comprehensive but unfocused?
   - A. It is annoying
   - B. It is acceptable if complete
   - C. It depends on the structure
   - D. It is fine if easy to read
4. Which reasoning path do you prefer?
   - A. Top-down
   - B. Bottom-up
   - C. Mostly top-down with supporting evidence
   - D. Mostly bottom-up with later synthesis
5. When learning a new concept, what do you want first?
   - A. What it is
   - B. How it differs from nearby concepts
   - C. What it is useful for
   - D. What mechanism drives it
6. What should happen when the problem is ambiguous?
   - A. Decompose it into subproblems
   - B. Give a best-effort answer
   - C. Offer a few interpretations to choose from
   - D. Point out where the ambiguity is
7. What makes an answer convincing?
   - A. Clear logic and reasoning
   - B. Evidence or authority
   - C. Actionability
   - D. Explaining why alternatives are weaker
8. When multiple options exist, what do you want first?
   - A. The best option
   - B. Fit by situation
   - C. One recommendation plus alternatives
   - D. Decision criteria
9. How should theory-heavy topics be handled?
   - A. Translate into practice quickly
   - B. Preserve theoretical integrity
   - C. Theory first, then application
   - D. Application first, then theory
10. How much implicit-intent completion do you accept?
   - A. High, do it by default
   - B. Medium, but label assumptions
   - C. Low, avoid it
   - D. It depends on the task

## Round 2: Work Style

1. Where do you naturally start a task?
   - A. Final deliverable
   - B. Gather context first
   - C. Start small and think while doing
   - D. Clarify what the stakeholder really wants
2. What matters most in progress?
   - A. Closer to the goal
   - B. More complete information
   - C. Better-crafted solution
   - D. Smoother process
3. Which builder style sounds most like you?
   - A. Build the frame first
   - B. Ship a rough version, then iterate
   - C. Eliminate key risks first
   - D. Hold the result and keep correcting course
4. What ambiguity annoys you most?
   - A. The goal is unclear
   - B. The boundaries are unclear
   - C. The priority is unclear
   - D. Ownership is unclear
5. What should a good solution satisfy first?
   - A. It can land in reality
   - B. It is logically rigorous
   - C. It is easy to communicate
   - D. It is reusable
6. How do you feel about MVP-first work?
   - A. Strongly agree
   - B. Agree if the direction is right
   - C. Neutral
   - D. Usually prefer fuller planning
7. Which role do you play most often?
   - A. Translate abstract goals into tasks
   - B. Organize messy information
   - C. Bridge research, product, and technology
   - D. Guard the outcome and adjust direction
8. What wastes time most?
   - A. Endless discussion without output
   - B. A lot of work that does not matter
   - C. Rework caused by weak early thinking
   - D. Activity without closure
9. Which work pattern earns your trust?
   - A. Clear deliverables and success criteria
   - B. Extensive analysis and discussion
   - C. Milestones and stage validation
   - D. A strong owner making calls
10. What role should the assistant default to?
   - A. Problem definition and goal calibration
   - B. Structured organizer and decomposer
   - C. Draft-and-iterate partner
   - D. Decision support and prioritization advisor

## Round 3: Output Preference

1. What is the best opening?
   - A. One-line conclusion
   - B. Restate the real problem first
   - C. Explain why the topic matters first
   - D. Preview the answer structure first
2. What answer length do you prefer?
   - A. As short as possible
   - B. Medium by default
   - C. Slightly longer but clearly structured
   - D. Long is fine if dense and useful
3. Which organization style works best?
   - A. Flat bullet points
   - B. Headings plus paragraphs
   - C. Table
   - D. Tree or framework
4. What should the answer optimize for first?
   - A. Immediate grasp of the main point
   - B. Logical completeness
   - C. Easy forwarding to others
   - D. Smooth reading flow
5. How should tables be used?
   - A. Use them whenever possible
   - B. Use them for comparison or classification only
   - C. Neutral
   - D. Avoid them unless necessary
6. What tone fits best?
   - A. Direct
   - B. Professional and rigorous
   - C. Peer-to-peer
   - D. Consultant-style
7. How should uncertainty be handled?
   - A. Give the most likely judgment and move on
   - B. Mark what is certain and uncertain
   - C. Give a main judgment plus risks
   - D. List multiple possibilities
8. Where should action guidance go?
   - A. At the start
   - B. At the end
   - C. Right after the conclusion
   - D. In its own explicit block
9. What makes writing readable for you?
   - A. Shortness
   - B. Clarity
   - C. Layered structure
   - D. Strong information density
10. If the assistant thinks your framing is off, how should it respond?
   - A. Say it directly
   - B. Affirm first, then say it
   - C. Restate your goal, then point out the mismatch
   - D. Offer an alternate wording and let you compare

## Round 4: Value Function

1. In important choices, what matters first?
   - A. Long-term value
   - B. Immediate results
   - C. More freedom
   - D. Stability
2. What kills your interest fastest?
   - A. No growth
   - B. No future or wrong direction
   - C. Bad management or collaboration
   - D. No real impact
3. What do you most want to invest time in?
   - A. Compounding capabilities and assets
   - B. Fast visible wins
   - C. A direction worth going deep into
   - D. Opportunities with strong external upside
4. How would you like to be known?
   - A. Insightful
   - B. Able to get things done
   - C. System-capable
   - D. Strong judgment
5. If an opportunity looks good but feels directionally wrong, what do you do?
   - A. Skip it
   - B. Test it on a small scale
   - C. Gather more information
   - D. Check market or peer signals
6. Where does the strongest sense of achievement come from?
   - A. Seeing the essence of a complex problem
   - B. Turning ambiguity into a clear artifact
   - C. Building a unique capability mix
   - D. Causing real impact
7. Which state do you most want to avoid?
   - A. Busy without accumulation
   - B. Stable without growth
   - C. Ideas without landing
   - D. Ability without direction
8. Which value is most central if you must choose one?
   - A. Growth
   - B. Impact
   - C. Autonomy
   - D. Correctness
9. What argument persuades you most?
   - A. This is the better long-term path
   - B. This moves the result now
   - C. This builds a rare moat
   - D. This matches the real problem and first principles
10. What risk do you watch most closely?
   - A. Choosing a dead-end direction
   - B. Spending effort without output
   - C. Being consumed by a weak organization
   - D. Failing to build core personal value

## Round 5: Collaboration Protocol

1. When the user's question is broad, what should happen first?
   - A. Decompose and prioritize
   - B. Provide a high-level frame
   - C. Give a best-effort answer
   - D. Identify the key clarification points
2. If the assistant thinks the user is asking the wrong question, what should it do?
   - A. Say so directly
   - B. Restate the real goal, then point out the mismatch
   - C. Answer first, then suggest a better framing
   - D. Only intervene if the mismatch is large
3. If information is incomplete, what should the assistant do?
   - A. Make reasonable assumptions and continue
   - B. List missing information and why it matters
   - C. Give a main judgment with explicit assumptions
   - D. Stay conservative and avoid judgment
4. When several valid answers exist, what should come first?
   - A. One recommendation
   - B. One recommendation plus alternatives
   - C. Decision criteria, then options
   - D. A full menu for the user to choose from
5. How much assistant-side convergence is acceptable?
   - A. High by default
   - B. High if the basis is explained
   - C. Medium and mostly for task work
   - D. Low unless requested
6. Which content split is most useful?
   - A. Fact / inference / recommendation
   - B. Conclusion / reasoning / action
   - C. Situation / problem / suggestion
   - D. Goal / obstacles / path
7. If the user provides mature wording, what should the assistant do?
   - A. Preserve it by default
   - B. Improve carefully while keeping intent
   - C. Rewrite boldly if stronger
   - D. First judge whether it is truly mature
8. What should plans or solutions usually include?
   - A. Risks
   - B. Limitations
   - C. Next-step actions
   - D. Alternatives
9. When should the assistant surface a higher-level stuck point?
   - A. Whenever it sees one
   - B. During hard decisions or repeated loops
   - C. Only when the user is clearly off-track
   - D. Rarely
10. What should the assistant remember long term?
   - A. Stable workstyle and communication preferences
   - B. Ongoing projects and context
   - C. Decision habits and value ordering
   - D. All of the above, with stable traits first

## Conflict Calibration

Use this round only when the earlier answers pull in incompatible directions.

1. If speed and completeness conflict, what is the default?
   - A. Get to a usable answer fast
   - B. Preserve completeness first
   - C. Get the direction right first, then move fast
   - D. It depends on the task
2. If autonomy and precision conflict, what is the default?
   - A. Make assumptions and continue
   - B. Continue only with labeled assumptions
   - C. Ask before assuming
   - D. It depends on task stakes
3. If shortness and context conflict, what is the default?
   - A. Be concise
   - B. Include the needed context
   - C. Start concise and expand when useful
   - D. It depends on whether the user is exploring or executing
4. If recommendation and criteria conflict, what is the default?
   - A. Recommend directly
   - B. Give criteria first
   - C. Give criteria plus one best-fit recommendation
   - D. It depends on how reversible the decision is
5. If rewriting and preserving wording conflict, what is the default?
   - A. Preserve wording
   - B. Improve lightly
   - C. Rewrite for clarity
   - D. It depends on whether the text is already mature
6. If one file and three files are both useful, what is the default?
   - A. One master profile
   - B. Split files
   - C. Master profile first, split only if requested
   - D. It depends on the target agent