# perception-first-design

Walks a page, email or pitch through five perceptual layers and tells you which one to fix first. Useful when the argument is about taste instead of causes.

- plugin · `type=plugin` · Claude Code · ★ 157
- Author: skovalik
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/skovalik/perception-first-design
- Updated: 2026-06-25
- HTML: https://affpapa.org/en/nevibecode/plugins/perception-first-design

## What it does

The plugin ships a skill that runs any material through five layers: cognitive load, the 50ms first impression, processing fluency, perception bias and decision architecture. You fix them bottom-up, since a failure low in the stack blocks everything above it. The framework itself runs 700+ lines with roughly 100 peer-reviewed citations.

Three modes. `evaluate` audits something that already exists — a URL, screenshot, mockup, HTML file or copy. `solve` derives a solution from cognitive constraints rather than instinct. `analyze` works through hypotheticals like removing the URL bar from a browser. A composite `all` mode runs all three on the same input. Without an explicit call, the skill picks a mode from how you phrase the request.

Match the number of passes to the stakes: the author suggests one pass for a quick design decision, five for a feature spec, ten for a landing page or pitch deck. Later passes return less than earlier ones, so you stop once confidence flattens out. Evaluation needs the artifact itself — a link, a file or text.

## Install

### Claude Code (native platform)

**Plugin marketplace**
In the agent chat (not a shell):
```text
/plugin marketplace add anthropics/claude-plugins-community
```
In the agent chat (not a shell):
```text
/plugin install perception-first-design@claude-community
```

## When to use it
- find out why visitors leave a landing page before reaching the headline
- derive interface requirements from cognitive constraints instead of taste
- show a stakeholder what is wrong with their instinct, citing research
- audit an onboarding flow or email for breaks in the path to action
- reason through a hypothetical like removing a UI element, layer by layer

## Example prompts
- PFD this landing page: https://example.com
- Run PFD on our onboarding flow design problem

## Similar
- [example-skills](https://affpapa.org/en/nevibecode/plugins/example-skills.md) — plugin · ★ 183839
- [impeccable](https://affpapa.org/en/nevibecode/plugins/impeccable.md) — plugin · ★ 88212
- [hyperframes](https://affpapa.org/en/nevibecode/plugins/hyperframes.md) — plugin · ★ 72481
- [playground](https://affpapa.org/en/nevibecode/plugins/playground.md) — plugin · ★ 44635
- [frontend-design](https://affpapa.org/en/nevibecode/plugins/frontend-design.md) — plugin · ★ 44635
- [visual-explainer](https://affpapa.org/en/nevibecode/plugins/visual-explainer.md) — plugin · ★ 10630

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md