Art — Visual Content Skill

What it does

The Art skill provides a complete visual content system for agents to produce illustrations, editorial headers, diagrams, and image edits. It implements aesthetic routing (branding-aware base prompt prefixes), workflow routing for types like mermaid flowcharts, technical diagrams, and sketchnotes, and model selection guidance (nano-banana-2 / nano-banana-pro). Concrete outputs include preview workflows, size and platform constraints, and pre-flight checks to avoid common diagram pitfalls.

When to use it

Use this skill when an agent must produce any web visual: blog headers, social cards, flowcharts, architecture diagrams, or iterative image edits. It triggers on requests mentioning art, header images, visualizations, mermaid/flowchart, infographic, or when the agent needs to pick an aesthetic for site-specific assets. Also useful when the agent should enforce brand style guides or follow platform size limits.

What's included

• Scripts: none packaged with the repo, but the skill documents CLI snippets for image generation and preview workflows.
• References: repository includes aesthetic.md, nano-banana-guide.md, and workflow templates for mermaid, visualise, and image-editing.
• Instructions: detailed rules for aesthetic selection, base prompt prefixes, model selection, and platform constraints to ensure consistent output.

Compatible agents

Designed for Claude/Claude Code workflows and image-generation pipelines that can call Gemini-like models (nano-banana-2/pro) and orchestrate prompt prefixes. It can be adapted by other agent runtimes that support executing the documented CLI patterns.