Skip to content

Repository files navigation

Image Skillbook — AI Image Skills for Codex & GPT Image

Turn real photos into doodle stories, crayon art, editorial paintings, and tactile keepsakes.

Image Skillbook is an open-source collection of seven independently installable Codex image Skills and one optional style router. Use them with GPT Image to transform a photograph into a tiny doodle story, crayon postcard, enamel keepsake, rubber-stamp journal, editorial painting, or impasto miniature world.

This is not a prompt archive. Every visual Skill packages source-aware adaptation rules, real before-and-after examples, one-command installation, and review checks.

Repository verification status [![License](https://img.shields.io/badge/license-MIT-2A9D8F)](LICENSE)

简体中文

Continued as Albert's Imagebook

Image Skillbook helped validate the first seven styles as independently installable Skills. Those packages and their existing install commands remain available. As the library grew, however, separate installs plus an extra selection router made one creative collection feel unnecessarily fragmented.

Active development and new styles have moved to Albert's Imagebook: one installable Skill, one $albert-imagebook command, and one growing style library.

npx --yes skills@1.5.24 add AlbertAZ1992/albert-imagebook \
  --skill albert-imagebook --global --agent codex --yes

This repository is now in maintenance mode. It will preserve the original Skill IDs and accept compatibility, security and documentation fixes, while new visual styles are added to Albert's Imagebook.

What Is an AI Image Skill?

An AI image Skill is an installable instruction package that tells a compatible agent how to adapt, run, and evaluate one repeatable visual treatment. Image Skillbook gives every style its own source-preservation rules, failure boundaries, real evaluation gallery, and exact install and invocation commands.

Install a Codex Image Skill

Install one treatment:

npx skills add AlbertAZ1992/image-skillbook \
  --skill photo-doodle-story --global --agent codex --yes

Start a new Codex session, attach a photograph, and mention the Skill by name:

Use $photo-doodle-story on this photograph.

Open a style below before installing it to see the source image, generated result, intended use, and invocation examples.

Find an AI Photo Style by Outcome

What you want to make Installable Skill
Add tiny doodle characters to a real photo $photo-doodle-story
Turn a photo into warm crayon art $crayon-memory-postcard
Make an enamel travel keepsake from a photo $enamel-travel-keepsake
Turn architecture or a landscape into rubber-stamp art $rubber-stamp-travel-journal
Repaint a photo as a sparse editorial memory $editorial-painted-memory
Reconstruct a photo with broad gouache or acrylic brushwork $painted-editorial-reconstruction
Turn a photo into a sculptural impasto miniature $impasto-miniature-world

See the Results

This comparison uses one source photograph so you can judge the treatments directly. Every result shown here was generated and reviewed.

Source photograph of a city skyline under a coral sunset

Editorial Painted Memory Enamel Travel Keepsake
City sunset reinterpreted as a sparse painted memory City sunset paired with a small enamel keepsake
Quiet acrylic shapes, restrained color, and generous paper. A source-derived enamel emblem paired with the authentic photo.
Crayon Memory Postcard Rubber Stamp Travel Journal
City sunset simplified into a warm wax crayon memory City sunset beside an aged paper carved stamp
A soft wax-grain recollection reduced to a few memorable elements. A field-journal spread with tactile broken ink.
Painted Editorial Reconstruction Photo Doodle Story
City sunset reconstructed with broad painterly brushwork City sunset turned into a miniature line-drawn story
A calm, airy retelling with broad visible brushwork. Source-locked photographic cues become a tiny line-drawn narrative.
Impasto Miniature World
City sunset sculpted into an impasto miniature world
A vivid, dreamlike environment sculpted from thick directional oil paint.

Full Evaluation Matrix

Every current visual Skill was run against the same four photographs: a skyline, a dense social space with signage, a strongly backlit beach, and a vertical tree-lined street. The accepted results below passed human review and moved the Skills to Candidate. Open any Skill page to see each source beside its result at a larger size.

Evaluation inputs

City sunset Coffee terrace Beach sunset Tree street
City skyline at sunset Busy coffee terrace People at a beach sunset Tree-lined historic street

Results

Treatment City Coffee Beach Street
Editorial Painted Memory Editorial painted city sunset Editorial painted coffee terrace Editorial painted beach sunset Editorial painted tree street
Enamel Travel Keepsake City sunset enamel keepsake Coffee terrace enamel keepsake Beach sunset enamel keepsake Tree street enamel keepsake
Crayon Memory Postcard City sunset crayon memory Coffee terrace crayon memory Beach sunset crayon memory Tree street crayon memory
Rubber Stamp Travel Journal City sunset rubber stamp Coffee terrace rubber stamp Beach sunset rubber stamp Tree street rubber stamp
Painted Editorial Reconstruction Painted city sunset reconstruction Painted coffee terrace reconstruction Painted beach sunset reconstruction Painted tree street reconstruction
Photo Doodle Story City sunset photo doodle story Coffee terrace photo doodle story Beach sunset photo doodle story Tree street photo doodle story
Impasto Miniature World City sunset impasto world Coffee terrace impasto world Beach sunset impasto world Tree street impasto world

Install Every AI Image Skill

Each style is independently installable. Install only the treatment you want, or install the whole collection:

npx skills add AlbertAZ1992/image-skillbook \
  --skill '*' --global --agent codex --yes

Need Help Choosing?

The optional $image-skillbook router compares the available treatments and gives you the install command for its recommendation.

npx skills add AlbertAZ1992/image-skillbook \
  --skill image-skillbook --global --agent codex --yes
Use $image-skillbook to recommend up to three treatments for this photo.
Explain the trade-offs and do not generate yet.

Requirements

These Skills require a Skill-compatible agent and an image generation or image editing tool. They are designed first for Codex with GPT Image, while their Markdown contracts remain portable to other compatible agents. Generated results vary by model, input, and instruction.

Maintenance Policy

Compatibility, security and documentation fixes for the seven independent Skills are welcome. Add new visual styles to Albert's Imagebook, where a style joins the shared workflow without becoming another installable SKU. The historical authoring process remains in CONTRIBUTING.md for maintainers of these packages.

Verify a Checkout

npm run check

This validates the catalog, Skill packages, documentation links, examples, and compatibility with the skills installer.

License

Repository-authored material is available under the MIT License. Adapted material must record compatible source rights before publication. Example provenance is documented in assets/examples/README.md.

About

Open-source AI image Skills and GPT Image prompt recipes for Codex—turn photos into doodle stories, crayon art, enamel keepsakes, rubber stamps, editorial paintings, and impasto worlds.

Topics

Resources

Contributing

Stars

3 stars

Watchers

0 watching

Forks

Releases

Packages

Used by

Contributors

Languages