Craft Skills
Five image-generation skills: Qwen-Image prompts, logos with shared geometry, diary comics, transparent PNGs and process dioramas
- 241 ★ on the repository
- Last commit
- Apache-2.0
Minimal access
Why this level
- The skills are instructions and check scripts, and your own model generates images
- Check scripts run on your machine and need Python
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/craft-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skills are instructions and check scripts, and your own model generates images
- Check scripts run on your machine and need Python
This is third-party code. Review the repository files before installing.
What it does
Craft Skills is a collection that turns professional methods into workflows with trigger boundaries, exit conditions, original examples and forward tests. The repository, written in Chinese with an English version, is not a pile of video summaries or prompt templates. Five skills ship. `qwen-image-gen` rewrites prompts for Qwen-Image, makes local edits and same-character series, adapts sizes and accepts results against the source image. `logo-semantic-fusion` designs and reviews logos whose several meanings share one geometry. `recurring-character-diary-comic` makes a 4–8 panel diary comic around an existing, authorized character: it compares three page skeletons and locks directions and verbatim dialogue. `native-transparent-imagegen` produces transparent PNG and WebP and checks the untouched alpha channel and thin edges, treating cutout faking as failure. `single-path-process-diorama` folds a 3–5 step main line into one miniature scene. Skills install by copying the folder you need into the agent's skills directory, for example `~/.codex/skills/`.
Who it is for. Designers and content creators who make images with a local Qwen-Image model and other generators and want a repeatable process instead of a one-off prompt.
Good fit when
- Qwen-Image is deployed locally but results miss the intent and resist edits
- You need a portrait series of one character with a consistent look
- A logo has to fuse two meanings in one shared shape
- You need transparent PNG or WebP with no checkerboard and no cutout faking
- A multi-step process must be shown as a single miniature scene
Not a fit when
- You need an image generator: the skills write and check prompts and process while your model draws the picture
- You want a comic about someone else's character without rights: the skill expects an existing, authorized character
- You want a quick one-off prompt with no checks: the set is built on process and acceptance
Example request
Rewrite this request for Qwen-Image with the qwen-image-gen skill and fit the size to a vertical coverLimitations
The skills are described mostly in Chinese, although an English version of the docs exists, and are aimed at a local Qwen-Image and similar generators rather than a cloud service. Some need Python and dependencies from `requirements.txt`, installed with a separate `pip install`. Each folder is copied separately, and the `scripts`, `references` and `assets` subfolders must travel with the skill. Skill status in the table varies: some carry versions below 1.0. The last commit is from September 2026.
How to disable. Delete the copied skill folder from the agent's skills directory, for example ~/.codex/skills/. The cloned repository can be removed separately.
FAQ
Which skills are included?
Five: qwen-image-gen, logo-semantic-fusion, recurring-character-diary-comic, native-transparent-imagegen and single-path-process-diorama.
How do I install one skill?
Clone the repository and copy the needed skill folder from skills/ into the agent's skills directory, for example ${CODEX_HOME:-$HOME/.codex}/skills/.
How does the transparent image skill differ?
It creates transparency natively and checks the untouched alpha channel and thin edges. An RGB checkerboard instead of transparency fails, and background cutout is not accepted.
What is recurring-character-diary-comic for?
A 4–8 panel diary comic about an existing character you have rights to: it compares three page skeletons and locks directions and verbatim dialogue.
Is there English documentation?
Yes, README.en.md sits next to the main README.
Discussion
No questions or reviews yet.
Related
SkillsA set of skills for Claude Code and Codex that assemble detailed image prompts from an element library and composition schemes
A set of SKILL.md skills for short-video and content production: topic, script, voice, avatar, editing, covers and motion graphics
Vane GPT Image Skills: two Codex skills for GIF sticker packs from a photo and pixel covers with illustrations for X articles
Qwen-Image 2.1 Prompter
qwen-image-2.1-skill
A skill that rewrites prompts for Qwen-Image-2.1 by Alibaba's official rules: text-to-image and multi-image editing
The skill turns paper text, methods and captions into structured prompts for scientific figures, tuned to the chosen image model.
Two skills for generative video and images: video writes Seedance, Kling and Veo prompts by dramaturgy rules, image picks the model and its syntax
README in short
The Chinese description presents the repository as a set of skills with trigger boundaries, exit conditions, examples and forward tests, then lists five skills in a table, shows qwen-image-gen samples and covers install and release checking.