# Pi Troupe recipes

## 1. Roast my README

```text
/troupe demo roast-my-readme
```

What to provide:

- Put `README.md` in current project root, or paste README/landing-page copy when prompted.

Expected report:

- above-the-fold critique
- confusing claims
- missing proof
- quotes from impatient user, skeptic, PM, maintainer
- concrete README edits and launch checklist

Best before:

- GitHub launch
- Product Hunt post
- npm release

## 2. Pricing survey

```text
/troupe demo pricing-survey
```

Custom version:

```text
/troupe Run a survey for my AI devtool pricing. Segments: solo developers, startup teams, enterprise leads. Questions: Would you try it this week? What blocks adoption? What would you pay monthly? Show segment breakdown and objections.
```

Expected report:

- synthetic sample assumptions
- aggregate distribution
- segment differences
- representative quotes
- next real survey questions

## 3. Landing page focus group

```text
/troupe demo landing-page-focus-group
```

Custom version:

```text
/troupe Run a focus group on this landing-page copy: <paste copy>. Find trust gaps, strongest promise, weak CTA, objections, and rewrite suggestions.
```

Expected report:

- persona-by-persona reactions
- disagreement transcript
- headline/CTA rewrites
- proof needed before conversion

## 4. Ad creative test

```text
/troupe demo ad-creative-test
```

Structured tool version with image:

```text
Run troupe_simulate with kind market_research, scenario "Evaluate this ad creative", imagePaths ["./ad.png"], exportResponses true.
```

Expected report: hook clarity, trust, CTA, objections, ratings, quotes, response CSV.

## 5. Screenshot usability test

```text
/troupe demo screenshot-usability-test
```

Structured tool version with image:

```text
Run troupe_simulate with kind software_testing, scenario "First-click UX test for this UI", imagePaths ["./screen.png"], exportResponses true.
```

Expected report: first-click targets, confusion, visual hierarchy, task-success risks, UX fixes.

## 6. Product Hunt launch roast

```text
/troupe demo product-hunt-launch-roast
```

Expected report: tagline issues, gallery gaps, first-comment rewrite, share triggers.

## 7. Feature prioritization

```text
/troupe demo feature-prioritization
```

Use when roadmap feels fuzzy. Output: divergent ideas, critique, ranked bets.

## 8. Onboarding UX test

```text
/troupe demo onboarding-ux-test
```

Use before release. Output: first-run confusion, edge cases, test scenarios.

## Guided UI

```text
/troupe
```

The wizard starts with clear simulation areas and descriptions:

- Survey — collect synthetic response rows
- Focus group — hear personas debate
- Interview — ask deeper follow-up questions
- Screenshot/UI review — evaluate a visual screen
- Ad / market research — test message or concept
- Brainstorm — generate and critique ideas
- Persona validation — check if personas are useful
- Small world — simulate a situation over time
- Custom — agent helps you shape the simulation conversationally

Then it asks only what matters:

- pasted text, image path, audience, or concept
- exact output artifacts: Markdown, CSV + Markdown, JSON + Markdown
- optional customization for objective/context/rounds

If you choose **Custom**, Pi Troupe switches to conversational setup: it asks follow-up questions, chooses the right simulation settings, and runs when enough information is available.

After run:

```text
/troupe logs
/troupe share
/troupe export csv
```

## Useful inspection commands

```text
/troupe templates
/troupe status
/troupe logs
/troupe share
/troupe export csv
/troupe export jsonl
```

Full-export surveys also write `survey-manifest.json` with batch status and `completedResponses / totalResponses`.