Structured framework for gathering requirements through interactive questionnaires. Based on the Conductor methodology's proven human-in-the-loop patterns.
When to Use
Setting up new projects
Defining product requirements
Gathering feature specifications
Onboarding users to new workflows
Any task requiring structured user input
Core Principles
1. Question Classification
Before asking ANY question, classify its type:
Type
Purpose
Phrasing
Example
Additive
Brainstorming, multiple answers valid
"Select all that apply"
"Which features do you need?"
Exclusive
Single choice required
No multi-select phrase
"Which framework should we use?"
2. Question Structure
All questions MUST follow this structure:
[Question text]
A) [Option A - often recommended, marked with "(Recommended)"]
B) [Option B]
C) [Option C]
D) Type your own answer
E) Auto-generate and continue
3. Sequential Questioning
CRITICAL
Ask ONE question at a time. Wait for response before next question.
CORRECT:
1. Ask Question 1
2. Wait for response
3. Confirm understanding
4. Ask Question 2
INCORRECT:
1. Ask Questions 1, 2, and 3 together
Questionnaire Workflow
Step 1: Introduction
Announce the section you're working on:
"I'll now help you define [section name]. I'll ask a few questions to understand your needs."
Step 2: Sequential Questions
For each question:
Classify
Is this Additive or Exclusive?
Formulate
Create clear question with options
Present
Show options in A/B/C/D/E format
Wait
Do NOT proceed without response
Confirm
Summarize understanding before continuing
Step 3: Handle Special Options
Option D (Type your own)
:
Accept user's custom input
Confirm the custom response
Continue to next question
Option E (Auto-generate)
:
Stop asking questions for this section
Use best judgment based on previous answers
Generate content and present for review
Step 4: Generate Content
CRITICAL
Use ONLY the user's selected answers as source of truth.
CORRECT:
- User selected "OAuth 2.0" -> Generate OAuth implementation details
INCORRECT:
- Include Option A, B, C text that wasn't selected
- Include question text in generated content
Step 5: User Confirmation Loop
After generating content:
"I've drafted [document name]. Please review:"
[Content preview]
What would you like to do?
A) Approve - proceed to next step
B) Suggest Changes - tell me what to modify
Please respond with A or B.
Loop until approved or user provides alternative direction.
Question Patterns
Product Vision Questions (Additive)
What are the primary goals of this product? (Select all that apply)
A) Improve user productivity
B) Reduce operational costs
C) Enhance customer experience
D) Type your own answer
E) Auto-generate and continue
Technology Selection Questions (Exclusive)
Which database should we use?
A) PostgreSQL (Recommended) - Best for relational data with complex queries
B) MongoDB - Document store for flexible schemas
C) SQLite - Lightweight for small applications
D) Type your own answer
E) Auto-generate and continue
Feature Scoping Questions (Additive)
Which authentication methods should be supported? (Select all that apply)
A) Email/Password
B) OAuth (Google, GitHub)
C) Magic Links
D) Type your own answer
E) Auto-generate and continue
Integration with AskUserQuestion Tool
When using the AskUserQuestion tool, map to this framework:
AskUserQuestion
(
{
questions
:
[
{
header
:
'Database'
,
question
:
'Which database should we use?'
,
multiSelect
:
false
,
// Exclusive question
options
:
[
{
label
:
'PostgreSQL (Recommended)'
,
description
:
'Best for relational data'
}
,
{
label
:
'MongoDB'
,
description
:
'Document store for flexible schemas'
}
,
{
label
:
'SQLite'
,
description
:
'Lightweight for small apps'
}
,
]
,
}
,
]
,
}
)
;
State Persistence
For multi-step questionnaires, save progress:
{
"last_successful_step"
:
"2.1_product_guide"
,
"answers"
:
{
"product_type"
:
"SaaS"
,
"target_users"
:
[
"developers"
,
"teams"
]
}
}
This enables resuming interrupted workflows.
Example Session
AGENT: "I'll help you define the product vision. First question:"
"Who are the primary users of this product?"
A) Developers
B) Business users
C) End consumers
D) Type your own answer
E) Auto-generate and continue
USER: A, B
AGENT: "Got it - targeting developers and business users. Next question:"
"What problem does this product solve?"
A) Streamlines workflows
B) Reduces manual work
C) Improves collaboration
D) Type your own answer
E) Auto-generate and continue
USER: D - "Automates code review feedback"
AGENT: "Understood - the product automates code review feedback for developers and business users. Let me generate the product vision..."
Anti-Patterns
Do NOT
Ask multiple questions at once
Include unselected options in generated content
Skip confirmation steps
Assume answers without asking
Use technical jargon in options without explanation
Do
One question at a time
Clear, concise option descriptions
Summarize understanding frequently
Include escape hatches (D, E options)
Respect user's custom inputs
Iron Laws
ALWAYS
ask exactly one question at a time and wait for a response before asking the next — presenting multiple questions simultaneously overwhelms users, produces ambiguous answers, and breaks the sequential state machine.
NEVER
use unselected option text in generated content — generated output must be built from the user's actual selected answers, not from the full list of options presented.
ALWAYS
classify each question as Additive (multi-select) or Exclusive (single choice) before asking — misclassification produces contradictory answers (user selects "B and D" when only one was valid).
NEVER
skip the confirmation step after generating content — presenting content without approval ignores refinements the user needed; always loop until the user explicitly approves.
ALWAYS
include a "Type your own" escape hatch option in every question — constrained option sets fail when the user's context doesn't fit any presented option; custom input prevents stalled workflows.
Anti-Patterns
Anti-Pattern
Why It Fails
Correct Approach
Presenting multiple questions at once
Ambiguous responses; breaks sequential state; users answer partially
Ask one question; wait for answer; then ask the next
Using option text verbatim in generated docs
Docs include options the user didn't choose; inaccurate requirements
Use only the selected answer values, not the full option list
Skipping question classification (Additive vs Exclusive)
Multi-select question treated as single choice or vice versa; contradictory requirements
Classify first; use
multiSelect: true
for Additive questions
Proceeding without confirmation
Generated content doesn't match user intent; rework required
Always present output for review; provide Approve / Suggest Changes loop
No custom input option
User's context doesn't fit any option; session stalls or forces wrong choice
Always include "D) Type your own" in every question