Skip to content
Back to skills

Recipe Refine Visuals

ASecurity

Optional side-workflow for design experts to refine auto-derived Visual Tokens in brand-direction.md with professional judgment

  • 10 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added May 27, 2026
data-aigo

Security analysis

A100/100

Scanned May 27, 2026

npx -y skills add shinpr/claude-code-discover --skill recipe-refine-visuals --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Recipe Refine Visuals?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Recipe Refine Visuals
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/shinpr-recipe-refine-visuals/badge)](https://www.skillsdirectory.com/skills/shinpr-recipe-refine-visuals)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: recipe-refine-visuals
description: Optional side-workflow for design experts to refine auto-derived Visual Tokens in brand-direction.md with professional judgment
disable-model-invocation: true
---

**Context**: Refine the Visual Tokens section of `docs/product/design/brand-direction.md` with design expertise. This recipe reads the existing brand direction (including auto-derived tokens from `recipe-blueprint`) and lets a design expert override specific values with professional judgment. The output is the same `brand-direction.md` file — downstream consumers (prototype-generator, UI Spec) read it without knowing which path produced the tokens.

## When to Use

- After `recipe-blueprint` has completed and `brand-direction.md` exists with auto-derived Visual Tokens
- When a design expert is available and wants to refine the visual system
- **This recipe is entirely optional** — auto-derived tokens are sufficient for prototype generation

## Orchestrator Definition

**Execution Protocol**:
1. **Follow the refinement flow** defined below
2. **Stop at every `[STOP — BLOCKING]` marker** — present findings and CANNOT proceed until user explicitly confirms

## Workflow Overview

```
Input (existing brand-direction.md with auto-derived tokens)
    ↓
1. Context Reading → Read brand direction + design principles + personas
    ↓
2. Token Review → Present current tokens with their derivation rationale
    ↓
3. Expert Refinement → User specifies overrides [Stop: User confirms changes]
    ↓
Output: Updated Visual Tokens in docs/product/design/brand-direction.md
```

## Execution Decision Flow

### 1. Context Reading

**Gate: `docs/product/design/brand-direction.md` must exist with a Visual Tokens section. If missing → inform user to run `recipe-blueprint` first.**

Read:
- `docs/product/design/brand-direction.md` — current direction and auto-derived tokens
- `docs/product/design-principles.md` — trade-off context
- `docs/product/personas/` — audience context
- Reference Products listed in brand-direction — use WebSearch to inspect their current design if needed

### 2. Token Review

Present the current Visual Tokens organized by category, showing:
- Current value and its derivation source
- How it relates to the directional sections above it
- Any inconsistencies between tokens (e.g., contrast ratio issues between text and surface colors)

Highlight areas where expert judgment would have the most impact:
- Color harmony and palette coherence
- Font pairing quality
- Spacing scale appropriateness for the product's density target

### 3. Expert Refinement

Guide the expert through each token category:

#### Color Tokens
- Present current palette as a visual summary (hex values with role labels)
- Ask: which values to keep, which to override
- Validate contrast ratios: `--color-text` against `--color-surface` must meet WCAG AA (4.5:1 for body text)
- Validate color harmony across the full palette

#### Typography Tokens
- Present current font selections with Google Fonts links
- Ask: alternative font pairings, size adjustments, weight preferences
- Validate: heading/body contrast is sufficient for hierarchy
- Validate: selected fonts support the product's language(s)

#### Spacing Tokens
- Present current scale with the base unit
- Ask: base unit adjustment, scale ratio preferences
- Validate: scale produces enough differentiation between levels

For each override, record the expert's rationale in the Decisions Log.

**[STOP — BLOCKING]** Present the refined token set to user for confirmation:
- Side-by-side comparison: auto-derived vs. expert-refined values
- Rationale for each change
- Contrast and harmony validation results

**CANNOT update the file until user explicitly confirms.**

### 4. File Update

After user approval:
- Update the Visual Tokens section in `docs/product/design/brand-direction.md`
- Change Source field from `auto-derived` to `expert-refined`
- Add override entries to the Brand Direction Decisions Log

## Scope Boundaries

**Included**: Visual Token refinement (colors, typography, spacing) within the existing brand-direction.md
**Not included**: Changing directional sections (Tone & Voice, Color Direction, etc.), adding new sections, component-level specifications, responsive breakpoints

## Completion Criteria

- [ ] Existing brand-direction.md read with auto-derived tokens
- [ ] Current tokens reviewed with derivation context
- [ ] Expert overrides collected with rationale
- [ ] Contrast and harmony validated
- [ ] User confirmed refined tokens
- [ ] brand-direction.md updated with source marked as `expert-refined`
- [ ] Decisions logged in Brand Direction Decisions Log

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.