Banner Design is a prose-only workflow for making banners across social media, display ads, website heroes and print. It asks for purpose, platform, content, brand and style, picks two or three art directions, and builds each as HTML and CSS at the exact pixel size.
A bundled reference lists platform sizes and 22 art styles. Safe zones, one call to action, font limits and contrast rules are built into the steps, and each option is captured as a PNG when the runtime can take screenshots.
What You Get
- Two or three banner options per request, built as HTML and CSS at the exact platform dimensions
- A size table covering Facebook, X, LinkedIn, YouTube, Instagram, Google Ads and website hero formats
- A reference of 22 art direction styles with the products each suits
- PNG exports named by style and size inside an assets/banners campaign folder
- Design rules for safe zones, one call to action, font count, contrast and print bleed
Common Use Cases
- Creating a LinkedIn or X header that matches a brand
- Producing display ad sizes for a campaign launch
- Designing a website hero visual with several style options
- Preparing a print banner with bleed and resolution notes
What Makes It Unique
It ships no scripts and no dependencies. Everything is a written workflow plus one reference file. When a capability such as screenshot capture is missing, the workflow says what to do: deliver the HTML source and mark PNG export as pending. The larger design skill in the same repository bundles banners with logos, icons and slides, and this one stays focused on banners only.
Under The Hood
The model reads SKILL.md and one reference file, then writes HTML and CSS files and, when the runtime offers a browser or screenshot tool, exports PNGs into an assets folder in the project. The skill itself runs no scripts and calls no services. Optional image generation or browser research happens only if the host runtime already provides them.