Refactor high-complexity React components in Dify frontend. Use when `pnpm analyze-component...
npx skills add nibzard/skills-marketplace --skill "brand-illustrator"
Install specific skill from multi-skill repository
# Description
Generate Builder Methods hand-drawn line art illustrations (icons, scenes, periphery) with a single accent color (Coral/Teal/Indigo/Amber). Use for blog headers, thumbnails, course graphics, social posts, and on-brand UI/tech metaphors.
# SKILL.md
name: brand-illustrator
description: Generate Builder Methods hand-drawn line art illustrations (icons, scenes, periphery) with a single accent color (Coral/Teal/Indigo/Amber). Use for blog headers, thumbnails, course graphics, social posts, and on-brand UI/tech metaphors.
allowed-tools:
- Bash
- Read
- Write
- Edit
Builder Methods Brand Illustrator
Generate hand-drawn line illustrations that match the Builder Methods visual identity:
warm off-white canvas, confident black ink lines, and one bold accent color.
This Skill is designed to be reusable and procedural:
- you gather requirements
- you propose three on-brand concepts
- you get a selection
- you generate final image(s) consistently using our style + color system
What this Skill produces
- Icon: 1 primary object, isolated, quick punctuation.
- Scene: 2–4 objects, suggested environment, hero/header moments.
- Periphery: 1–3 objects, used as corner/edge elements or decorative supports.
See references/visual-world.md for the world + constraints and references/style.md for rendering rules.
Requirements
- Python 3.8+ for running the generation script
- google-genai package:
pip install google-genai - GEMINI_API_KEY environment variable: Set your Google Gemini API key as
GEMINI_API_KEY(alternativelyGOOGLE_API_KEYorGENAI_API_KEY)
Get an API key from: https://ai.google.dev/
Quick Start
- Create a project folder (recommended) in
projects/with today's date and a short slug.
Example:
bash
mkdir -p projects/2026-01-13-blog-hero
- Gather requirements — use the AskUserQuestion tool for each missing piece one at a time.
Required inputs: - Content context: topic + the core idea (insight), or paste article/transcript
- Visual context: where the illustration will live (page screenshot, layout notes), or “n/a”
- Accent color:
coral|teal|indigo|amber - Image type:
icon|scene|periphery - Dimensions (px): width × height
Defaults:
- scene: 1200×630
- icon: 512×512
- periphery: 500×500
- Generate 3 concept options using
references/idea-mapping.md: - Present Option A / B / C
- Each option includes: connection type, category (Builder’s World / Metaphor / Digital Artifact),
object list, and why it fits the content. -
Keep options meaningfully different (object choice, metaphor, or mood), but all on-brand.
-
Get user choice — use AskUserQuestion and let them pick A/B/C (or “Other” for feedback).
-
Document the project — create
project.mdinside the project folder with: - requirements
- A/B/C concepts
- chosen direction
- final prompt + parameters
-
output filenames
-
Craft prompt and generate once a concept is approved, saving outputs to the project folder.
Example:
bash
python3 scripts/generate.py \
--prompt "A worn leather notebook open to a page with handwritten wireframe sketches" \
--color coral \
--type scene \
--width 1200 \
--height 630 \
--output projects/2026-01-13-blog-hero/illustration-v1.png
Color System
See references/colors.md for all hex values (single source of truth).
Available accent colors: Coral, Teal, Indigo, Amber
Rules
- Use ONE accent color per illustration.
- Most of the image is warm off-white background with confident black lines.
- Accent color should fill ~20–30% of the illustration.
- Shadow/depth color is used sparingly (~5–10%), mostly as grounding.
Style Requirements (non-negotiable)
These are enforced by prompt + review. See references/style.md for full details.
- Hand-drawn ink line art; no photorealism, no 3D, no gradients
- Restrained compositions: focus objects over scenery
- Suggest environment with a few cues; do not render full rooms
- Leave negative space for text overlays when used as a hero image
Concepting Guidance
Use the mapping doc to ensure every illustration is connected to the content:
- references/idea-mapping.md — connection types, object lists, metaphors, quick reference by meaning
- references/visual-world.md — what “belongs” in the Builder Methods world
- references/prompts.md — prompt templates and proven patterns
Safety / Brand Guardrails
- Avoid trademarks/logos/brand names on devices, mugs, screens, or apparel.
- Avoid depicting real identifiable people.
- Keep UI/terminal content generic (nonsense code is fine; no secrets).
- No violence, gore, or sensitive themes—Builder Methods illustrations should feel calm and inviting.
Project Documentation Template
Copy into projects/<date>-<slug>/project.md:
# Project: <slug>
## Requirements
- Content context:
- Core idea:
- Visual context:
- Accent color:
- Image type:
- Dimensions:
## Concepts
### Option A
- Connection type:
- Category:
- Objects:
- Rationale:
### Option B
...
### Option C
...
## Selected Direction
- Chosen option:
- Notes / tweaks:
## Final Prompt
```text
<final prompt here>
Generation Params
- color:
- type:
- width:
- height:
- output:
Outputs
- illustration-v1.png
- illustration-v2.png
```
# Supported AI Coding Agents
This skill is compatible with the SKILL.md standard and works with all major AI coding agents:
Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.