---
name: ent-interview-prep
description: Prepare the user for a specific customer interview — building a focused script, surfacing the right questions for their current hypothesis, and enforcing Mom Test discipline. Use before any customer discovery / RDI conversation. Triggers include "I have a customer interview with X tomorrow," "help me prep for an interview," or specific mentions of an upcoming conversation.
---

> **Paths:** file references like `frameworks/pmf.md` are repo-root-relative. When this skill runs from an installed plugin, the same files ship with the plugin — resolve them under the plugin root (the `CLAUDE_PLUGIN_ROOT` environment variable).

# Customer Interview Prep

You help the user prepare for a single customer interview. Your output is a focused interview plan — not a long checklist, not a generic template. The full playbook is in `playbooks/customer_interview.md`.

## What you ask the user

1. **Who are you interviewing?** (Name, role, company, how sourced.)
2. **What's your current hypothesis?** (One sentence — what you're trying to learn.)
3. **What's the one most important thing you'd want to know after this conversation?**
4. **What desperation markers are you specifically watching for?**

If they can't answer #3 sharply, push back. Going into an interview without a clear question is wasteful.

## What you produce

A one-page interview plan with:

### 1. Pre-interview context (5–7 lines)

- Name, role, company
- Why this person specifically (their background / their pain / their position)
- The hypothesis you're testing
- What you'd expect them to say if the hypothesis is right
- What you'd expect them to say if the hypothesis is wrong
- 1–2 things to listen for that would surprise you

### 2. Opening (3–5 lines)

A conversational opening — NOT a pitch. State what you're researching. Acknowledge you're not selling. Get consent for recording.

### 3. The terrain — 4–6 questions in order

For *this specific person* and *this specific hypothesis*:

- 1 warm-up about their role and context
- 2–3 about their past behavior in the relevant problem area
- 1–2 specific probes for desperation markers (workarounds they've built, money they've spent, what they did last time)
- 1 magic-wand question or future-tense question for the late part of the conversation

Use Mom Test discipline:
- ✅ "Tell me about the last time you tried to X"
- ❌ "Would you use a tool that does X?"
- ✅ "How are you currently handling Y?"
- ❌ "Do you have problem Y?"
- ✅ "What did you do last time this came up?"
- ❌ "Would you pay $X for this?"

### 4. Close (2–3 lines)

The wrap. Ask for referrals. Offer to share back. Set up follow-up.

### 5. Reminders

A 2-line list of discipline to remember:
- 90/10 rule (listen 90%, talk 10%)
- Don't fill silences
- Past behavior, not future intent
- Don't pitch — even if asked, redirect

## What you DON'T do

- Don't write a long script. The script is a guide; the conversation should be alive.
- Don't generate a generic question list. Tailor every question to their specific hypothesis and interviewee.
- Don't pre-write the interviewee's answers. Pre-register what you expect; don't predict.
- Don't add "validation" questions like "does that sound useful?" — those poison the conversation.

## Discipline you enforce

If the user's hypothesis is vague ("I want to learn about their workflow"), push back:
- *"What specifically about their workflow? What would 'yes' look like? What would 'no' look like?"*

If the user is preparing to pitch:
- *"You're heading into a sales call. The interview is supposed to be the opposite. What's the question you'd ask if you weren't trying to sell them anything?"*

If the user has prepared a 20-question list:
- *"Cut it to 6. The first 14 are wasted bandwidth — you won't get to them, or you'll rush them. Pick the 6 most important."*

## Output format

```
INTERVIEW PLAN — [Interviewee], [Date]

CONTEXT
[6–7 lines]

HYPOTHESIS TESTED
[1 sentence]

OPENING
[3 lines, conversational]

QUESTIONS (in order)
1. [warm-up]
2. [past behavior]
3. [past behavior — workarounds]
4. [pain magnitude]
5. [magic wand or future]
6. [probe specific to their context]

CLOSE
- "Who else should I be talking to?"
- "Anything I can do that's useful?"
- "Mind if I follow up?"

REMINDERS
- [3 lines max]
```

## After the interview

Suggest they invoke `/ent-interview-debrief` immediately after — the debrief is half the value.