GitHub radar
Make your GitHub project page clear at first glance
A skill for your AI coding agent turns a wall of text on your project's main page into a tidy page with images and a clear structure — no manual design work.
It's an instruction set for your AI coding agent (Claude Code, Codex, and similar) that redesigns your project's main GitHub page: decides what to show first and adds diagram-style images.
When it helps
Useful if your GitHub project page is a pile of commands nobody reads past the first screen. Not needed if you don't have a public GitHub project with such a description page.
Pros
- Picks a layout based on your project type, not one template for all
- Shows a local preview before anything gets published
- Doesn't fake stars, usage numbers, or endorsements for your project
- Asset-only mode leaves your text and links untouched
Cons
- Whole-page mode rewrites your text — read it before agreeing
- AI-generated images need a separate image generation tool
How to set it up — step by step
- 1Open your AI coding agent inside your GitHub project folder.
- 2Send the agent the command npx skills add oil-oil/beautify-github-readme to install the skill.
- 3Send the agent the text from the block below to set the task and mode.
- 4Choose asset-only mode if you don't want your text changed.
- 5Check the local preview of the result before publishing anything.
- 6Approve publishing separately, only if you like the result.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up the beautify-github-readme skill from https://github.com/oil-oil/beautify-github-readme — install it with npx skills add oil-oil/beautify-github-readme, run it in asset-only mode on my repository, and show me the local preview first. Do not publish anything without my approval, and ask me if you need access to an image generation tool.
This is not a standalone program. The skill installs into your AI coding agent through npx and needs no special hardware. If you want an AI-generated image like the Wolfcha example in the README, you will need separate access to an image generation tool.
Open on GitHub▌ More finds