distinctive-ai-design skill
A Codex skill: three distinct visual concepts, screenshot review and cutting excess, so interfaces look less templated.
- 135 ★ on the repository
- Last commit
Minimal access
Why this level
- The skill only sets the working order and runs nothing itself; the agent makes the project edits.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/distinctive-ai-designDetects 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 skill only sets the working order and runs nothing itself; the agent makes the project edits.
This is third-party code. Review the repository files before installing.
What it does
The skill runs a design task in three steps: explore, define, deliver. The agent first reads the project and proposes about three concepts that differ in composition, information structure, type and interaction metaphor, not just colour. The chosen one is condensed into a short brief and built on the current stack. The agent then captures screenshots of the running interface and reviews them: at most two rounds, each fixing one to three main problems. The last step is subtraction: remove repeated labels, empty containers and effects that compete with the primary action. The skill instructions are written in Chinese; tasks can be given in Russian or English.
Who it is for. Developers and designers in Codex who want an interface with character: a new landing page, a game screen, or a rework of a page that works but looks generic.
Good fit when
- You need a visual direction for a new product or a game menu.
- A working page looks templated and should be redone without changing features.
- A style is approved and is being refined across type, layout and interaction.
- The interface is crowded with labels and frames and needs a screenshot-driven cleanup.
Not a fit when
- A single margin or colour fix: the skill targets a whole design task.
- A brand book fixes every element and leaves no room for choice.
- An environment with no screenshots: there is nothing to base a visual review on.
Example request
Use $distinctive-ai-design for the home page. The user is an independent developer and the main action is starting a trial. Propose three different directions and build the best one on the current stack.Limitations
The skill brings no browser, image generation, API keys or separate review model. Without screenshots the agent marks visual verification as not done, and without an independent subagent the review becomes a self-check. Instructions are in Chinese, the README exists in English. It is based on an Anshu Chimala article in Lenny’s Newsletter: techniques 1–6 were read, the paid seventh was not. Last commit on 10 October 2026.
How to disable. Delete `~/.agents/skills/distinctive-ai-design` or `.agents/skills/distinctive-ai-design` in the project.
FAQ
Is the full process needed for a small fix?
No. Ordinary feature edits and local style fixes are done directly; the full cycle is for concepts and redesigns.
What is the random seed in the skill?
The agent takes a random string from a local source such as `python3 -c 'import secrets; print(secrets.token_hex(48))'` and draws rhythm and contrast ideas from it. The seed never appears in the interface.
How many review rounds does the skill run?
At most two. Each fixes the one to three most important problems, then new screenshots are taken.
How do I update the skill?
If it was cloned with git and has no local changes, run `git -C ~/.agents/skills/distinctive-ai-design pull --ff-only`.
Does the skill need separate services?
No. A browser, image generation and API keys are not bundled and are used only if the environment already has them.
Discussion
No questions or reviews yet.
Related
SkillsA UI design skill: art direction per product, a shared DESIGN.md, tokens, and layout and accessibility checks in a browser.
Four Codex skills covering product design stages, from reviewing a single screen to comparing visual directions.
An interface design skill for the agent: diagnosis, four directions to choose, implementation and QA without an AI-template look
A UI, presentation and graphics design skill for coding agents: canvas drafts driven through the superdesign CLI
A skill for interfaces without the generic AI look: typography, color, motion, component states and a critique of the result.
A registry of installable design skills: ready SKILL.md and DESIGN.md files with tokens, component rules and accessibility for different styles
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| distinctive-ai-design (this card) | no | 135 | |
| Design Judge Skills | no | 711 | judges a finished design; this one covers the whole process from three concepts to a screenshot cleanup |
| Taste Skill | no | 114,100 | breaks a site from a link into tokens and rules; this one drives your own design from scratch or a rework in the project |
README in short
The skill pins no model, framework or paid service. Image and video generation are used only when the environment offers them and they serve the design.