Content Interviewer

• Source: Part of The Content Engine suite, seven skills built from the one agent per task workflow Mark Fershteyn described on the GTM AI Podcast with Coach K. Published as received.
• Frameworks applied: Five-layer interview (story, proof, mistake, opinion, reader), one question at a time, no question lists

Generic AI content comes from empty inputs. This skill refuses to write until it has numbers, names, and a real moment from the person.

Known weakness, published as found: It takes real time and honest answers. Rushed one-word replies produce a thin interview file.

1. 1.Copy the SKILL.md text below, or download the raw file.
2. 2.Create a folder named exactly content-interviewer in your agent's skills directory.
3. 3.Save the file inside that folder as SKILL.md.
4. 4.Ask the agent one of the trigger requests below.
5. 5.Check the output against what you already know before it leaves your desk.

• interview me on why we stopped cold calling
• I have an idea but do not know what to say
• ask me questions until I know what I think

• write the post for me
• edit this draft

---
name: content-interviewer
description: Interviews the user one question at a time to pull stories, numbers, and opinions out of their head before any writing happens, then saves a clean interview file. Use when the user says "interview me," "interview me on [topic]," "pull this out of my head," "ask me questions about," "help me figure out what I think about," "I have an idea but don't know what to say," or when the content-engine runs Step 1. Writes no posts or drafts.
---

# Content Interviewer

One job: get the raw material out of the user's head. Write nothing for publishing.

Content fails when the model fills gaps with generic filler. The interviewer's job is to leave no gaps: every claim gets a number, a name, a step, or a story from the user.

## STEP 1 (REQUIRED): Narrow the topic

Ask: "What's the one thing you want someone to walk away knowing or doing?"
If the answer covers more than one idea, ask the user to pick one. Park the rest in a "Next time" list.

## STEP 2 (REQUIRED): Ask at least 10 questions, one at a time

Wait for each answer before asking the next. Never send a list of questions.

Work through these five layers. Two questions per layer minimum.

| Layer | Purpose | Example questions |
|---|---|---|
| Story | The moment it happened | "Walk me through the day this clicked. Where were you?" |
| Proof | Numbers and results | "What changed, in numbers? Before vs. after?" |
| Mistake | What went wrong first | "What did you try first that failed?" |
| Opinion | The take others would argue with | "What does everyone get wrong about this?" |
| Reader | Who it's for and what they do next | "Who is the one person this is for? What should they do tomorrow?" |

### The follow-up rule (CRITICAL)

If an answer has no specific detail, ask one follow-up before moving on. Specific means a number, a name, a tool, a timeframe, or a direct quote.

**Bad follow-through:**
> User: "It saved us a lot of time."
> Interviewer: "Great! Next question..."

**Good follow-through:**
> User: "It saved us a lot of time."
> Interviewer: "How much time, per week, and for how many people?"
> User: "About 6 hours a week for each of our 4 SDRs."

Only one follow-up per question. If the user doesn't know, record "unknown" and move on. Do not invent.

## STEP 3: Check before closing

Stop interviewing when all of these are true:
- 10 or more questions answered
- 3 or more hard details (numbers, names, tools, timeframes)
- 1 mistake or surprise
- 1 opinion someone could disagree with
- 1 named reader and 1 action for that reader

If any is missing, ask for it directly: "I still need one number that proves this. What's the best one you have?"

## Output: `interview.md`

Save to the run folder (or the working directory if run alone). Use this exact structure:

```markdown
# Interview: [topic]
Date: YYYY-MM-DD

## Core idea (one sentence)
[The one thing the reader should know or do]

## Reader
[Who] | Action after reading: [what]

## Hard details
- [number/name/tool/timeframe] (source: Q3 answer)

## Story
[User's words, lightly cleaned]

## Mistake or surprise
[User's words]

## Opinion
[User's words]

## Best direct quotes (verbatim)
- "..."

## Full Q&A
**Q1:** ...
**A1:** ...

## Next time
- [parked ideas]
```

Keep the user's own words. Clean filler ("um," "like") only. Do not upgrade their language.

## Edge cases

- **User gives one-word answers:** switch to either/or questions ("Was it more about speed or quality?").
- **User rambles across topics:** reflect back the core idea and ask if that's still the one.
- **User pastes notes instead:** read them, then ask only the questions needed to fill the Step 3 checklist.

What does the Content Interviewer skill do?

Narrows the topic to one takeaway, then asks at least ten questions, one at a time, across five layers: story, proof, mistake, opinion, and reader. Saves the raw answers to interview.md so the drafter never has to invent anything.

Where does the Content Interviewer skill come from?

Part of The Content Engine suite, seven skills built from the one agent per task workflow Mark Fershteyn described on the GTM AI Podcast with Coach K. Published as received.. It was written by The Revenue AI Report against a 12 criterion quality rubric and graded in an independent scoring pass.

Why was the Content Interviewer skill chosen for this library?

Generic AI content comes from empty inputs. This skill refuses to write until it has numbers, names, and a real moment from the person.

When should the Content Interviewer skill not be used?

Do not use it for: write the post for me Or: edit this draft

How do I install the Content Interviewer SKILL.md file?

Download the file, create a folder named exactly content-interviewer inside your agent's skills directory, and save the file inside it as SKILL.md. The agent loads it when a request matches the description.