skills.rest
edheltzel
edheltzelCOMMUNITY
Explore all skills by this author →
Website𝕏@edheltzelGitHub

do-art

Generates static visual content across 20+ formats using Flux, Nano Banana Pro, and GPT-Image-2.

2|Updated May 25, 2026
Search Tags:#image generation#diagrams#thumbnail#infographic#nano-banana-pro#gpt-image-2#blog header
One-click install
npx skills add https://github.com/edheltzel/Do-Skills --skill do-art-edheltzel
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: do-art
Source: https://github.com/edheltzel/Do-Skills/tree/main/skills/content/do-art
Command: npx skills add https://github.com/edheltzel/Do-Skills --skill do-art-edheltzel

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires rembg, imagemagick, and includes scripts (resource) and references (resource) components.

What problem does it solve? Freeform prompts to image models produce inconsistent, off-style output with wrong sizes, opaque backgrounds, and unreviewed files pushed straight into projects. This Skill enforces a fixed workflow-routing, model-selection, and Downloads-first review discipline in code so every generated image follows a defined technique and passes verification before use. ## Core Features & Use Cases - Workflow-gated generation: Every request routes through one of 20+ named workflow files (Essay, Mermaid, TechnicalDiagrams, Frameworks, D3Dashboards, Comics, Aphorisms, and more), and Generate.ts refuses to run without a --workflow flag. - Multi-model routing: Picks between Flux, Nano Banana Pro, and GPT-Image-2 based on the job, with per-model size and aspect-ratio validation. - Blog header pipeline: Produces both a transparent inline PNG and an opaque sepia social thumbnail via --thumbnail, with local rembg background removal and RGBA verification. - Use Case: Ask for a header image for a blog post and receive a workflow-composed editorial illustration staged in ~/Downloads/ with both transparent and social-thumbnail variants ready for review. ## Quick Start Ask the agent to create a header image for your blog post and it will follow the Essay workflow, generate the image to ~/Downloads/, and produce both inline and thumbnail versions for your review.

Frequently Asked Questions about do-art

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a blog header image with an AI image model?▼

Route the request through the Essay workflow, then call Generate.ts with --workflow=Essay, a model such as nano-banana-pro, and the --thumbnail flag. This produces both a transparent inline PNG and an opaque sepia social thumbnail staged in ~/Downloads/ for review.

Which image model should I use for text-heavy diagrams?▼

Use gpt-image-2 for text-heavy work like stat cards, framework diagrams, taxonomies, and timelines because it has the strongest text rendering. Use nano-banana-pro for editorial illustrations and flux for stylistic variety or technical illustration.

Why does Generate.ts refuse to run my image prompt?▼

Generate.ts exits non-zero unless you pass --workflow=<name> matching a file under Workflows/, or the explicit --freeform-confirmed opt-out. This gate exists to prevent freeform prompts that skip the composition and palette rules encoded in each workflow.

Why does my transparent PNG have a white or black background?▼

Background removal via rembg can fail on thin linework, and flattening an opaque JPEG onto a background color is a no-op. Verify with the file command that output reports RGBA, and for thin-line diagrams either prompt for thick saturated linework or skip background removal when the page background matches.

Can I use reference images for character consistency?▼

Yes, pass multiple --reference-image flags to Generate.ts for improved likeness and style consistency. The Gemini API allows up to 5 human reference images, 6 object references, and 14 total per request.

When should I not use this image generation skill?▼

Do not use it for locked house-style YouTube channel thumbnails, video or animation (use Hyperframes or Remotion instead), or web UI design and integrated frontend layout. It covers static visual content only.

Related Skills

Explore More Design & Creative Skills→
gooseworks-aigooseworks-ai
6

source-brand-assets

Scrapes logos and product photos from the web into a structured brand-assets folder using curl and web search.

Official
Intermediate
iterationlayeriterationlayer
2

generate-thumbnail

Resize source images to thumbnails and convert them to WebP.

Official
Intermediate
timescaletimescale

linkedin-article-writer

Convert blog drafts and Tiger Den items into LinkedIn articles with SEO metadata and social posts.

Official
Advanced
RooseveltAdvisorsRooseveltAdvisors
1

Media

Generate illustrations, diagrams, and programmatic videos with Remotion.

Community
Advanced