/interview

name: plan-interview:interview
description: Adaptive interview that probes plans with depth proportional to complexity, generating comprehensive specifications
argument-hint: [plan]
model: opus

Adaptive Plan Interview & Spec Generation

Read the plan file at `$ARGUMENTS` thoroughly. Before asking questions, assess the plan's complexity to calibrate interview depth. Then conduct a rigorous, multi-phase interview using `AskUserQuestion` to extract all implicit knowledge, assumptions, and decisions needed to write a complete specification.

---

Complexity Assessment

**Before starting the interview, classify the plan:**

| Complexity | Signals | Question Depth | |------------|---------|----------------| | **Simple** | Single feature, clear scope, minimal integrations, one user type | 10-15 questions total | | **Moderate** | Multi-component, some integrations, defined boundaries, 2-3 user types | 18-23 questions total | | **Complex** | Cross-system, many stakeholders, unclear boundaries, multiple user types, high risk | 22-28 questions total |

**Adapt dynamically**: If answers reveal hidden complexity, probe deeper. If an area is well-specified, move faster.

---

Interview Philosophy

**Core Principle**: Depth over breadth. Better to deeply understand critical aspects than superficially cover everything.

**Key Techniques**:

• **Ask non-obvious questions** — skip anything the plan already answers clearly
• **Probe edges and boundaries** — what happens in unusual cases?
• **Challenge assumptions** — "You mention X, but what if Y?"
• **Seek concrete examples** — "Give me a specific scenario where..."
• **Follow threads** — if an answer reveals complexity, dig deeper before moving on
• **Detect contradictions** — flag when answers don't align with previous statements
• **Surface assumptions** — make implicit beliefs explicit

> **Deep Dive**: Load `references/interview-techniques.md` for advanced probing patterns.

---

Interview Phases

Phase 1: Foundations & Scope (CRITICAL)

**Goal**: Establish crystal-clear understanding of who, what, why, and boundaries.

**Focus Areas**:

1. **Ambiguous Terminology**

• What terms could mean different things to different people?
• What concepts need explicit definition?

2. **Stakeholder Analysis**

• Who uses this? Who maintains it? Who is impacted?
• Who has veto power? Who must approve?
• Who will complain loudest if this fails?

3. **Success/Failure Criteria**

• How do we know this succeeded? What metrics matter?
• What does failure look like? What are early warning signs?
• What's the minimum viable success?

4. **Constraint Exploration**

• What CAN'T we do? (regulatory, technical, political)
• What's fixed vs negotiable?
• What resources are limited? (time, budget, people, expertise)

5. **Prior Art Investigation**

• Has this been tried before? Why did it fail/succeed?
• What similar solutions exist? Why not use them?
• What can we learn from others' mistakes?

6. **Dependency Mapping**

• What must exist before this can work?
• What depends on this completing?
• What external systems/teams are involved?

7. **MVP Scope Definition**

• What's the absolute minimum to prove value?
• What can be deferred without blocking progress?
• What features seem essential but aren't?

**Example Questions**:

• "Who will be most upset if this doesn't work? Why?"
• "What's one thing you're assuming is true but haven't verified?"
• "If you had to ship in half the time, what would you cut?"
• "What's the worst-case scenario if this fails?"

**Target**: 4-6 questions (simple: 2-3, complex: 5-6)

> **Deep Dive**: Load `references/phase-1-clarifications.md` for 15+ example questions and common pitfalls.

---

Phase 2: Technical Deep-Dive

**Goal**: Understand the technical approach, architecture, and engineering trade-offs.

**Focus Areas**:

1. **Architecture Decisions**

• What's the high-level structure?
• Why this architecture over alternatives?
• What patterns are we using/avoiding?

2. **Data Models & State**

• What are the key entities and relationships?
• Where does state live? How is it synchronized?
• What's the source of truth for each data type?

3. **APIs & Integrations**

• What external systems do we touch?
• What contracts must we honor?
• What happens when integrations fail?

4. **Scalability Exploration**

• What happens at 10x, 100x, 1000x scale?
• Where are the bottlenecks?
• What breaks first under load?

5. **Technical Debt Assessment**

• What compromises are we making? Why?
• What will we regret in 6 months?
• What's the cost of not fixing this later?

6. **Observability Planning**

• How do we know it's working in production?
• What metrics matter? What alerts trigger action?
• How do we debug issues at 3am with only logs?

7. **Migration Strategy**

• How do we get from current to new safely?
• What's the rollback plan if things go wrong?
• Data migration requirements?

8. **Security Considerations**

• What's the threat model?
• How is authentication/authorization handled?
• What data is sensitive? How is it protected?

**Example Questions**:

• "If this gets 100x more traffic tomorrow, what breaks first?"
• "What's the one technical decision you're least confident about?"
• "How would you debug this at 3am with only logs?"
• "What existing system will this integration disrupt?"

**Target**: 5-7 questions (simple: 3-4, complex: 6-7)

> **Deep Dive**: Load `references/phase-2-technical.md` for architecture discussion patterns.

---

Phase 3: User Experience

**Goal**: Understand how users will interact with and perceive the system.

**Focus Areas**:

1. **User Persona Depth**

• Who specifically uses this? (not just "users")
• How do novice vs expert needs differ?
• What do power users need that casual users don't?

2. **User Flows**

• What's the primary happy path?
• What are th