delight-audit
April 1, 2026 · View on GitHub
A Claude Code skill that audits Figma designs for emotional quality.
Most interfaces are competent. This skill finds the ones that feel — and fixes the ones that don't.
What it does
Point it at a Figma file and it evaluates every screen against three delight dimensions, then outputs a structured audit report with flagged flat moments and specific, implementable upgrade suggestions.
The three dimensions:
| Dimension | The question it asks |
|---|---|
| Unexpected joy | Does anything surprise the user in a good way? |
| Earned satisfaction | Does completing an action feel rewarding? |
| Emotional resonance | Does the screen feel alive, or does it just exist? |
For each screen you get a score (✓ / △ / ✗), an observation, and — for anything that's flat — a concrete suggestion for what to do about it.
Setup
- Install Claude Code
- Connect the Figma MCP server
- Add this skill to your Claude config:
{ "skills": ["path/to/delight-audit-figma/SKILL.md"] }
Usage
/delight-audit
Then provide your Figma file URL. Claude will read the file, evaluate each screen, and return the full audit report.
You can also scope it to a specific page or frame:
"Run a delight audit on the checkout flow in this file: [URL]"
Example output
### Checkout confirmation
| Dimension | Score | Observation |
|--------------------|-------|----------------------------------------------------------|
| Unexpected joy | ✗ | No moment of celebration after completing a purchase |
| Earned satisfaction| △ | Progress indicator exists but disappears abruptly |
| Emotional resonance| ✓ | Warm copy tone lands well — has a point of view |
Opportunity: Add a micro-animation on order confirmation — small, earned, joyful.
Runs once, doesn't repeat.
MCP tools used
get_design_context get_screenshot get_metadata
By Marie Spreitzer · MIT License