— How to use this repo

Make a page that looks like you.

A folder of VS Code Copilot skills that turn a mood board, or just a sentence, into a styled HTML presentation page.

Copy the skills
See examples

01 · install

Copy the skills folder into your own repo.

In your own repo, create .github/skills/ if it doesn't exist, then copy the three skill folders from this repo into it.

your-repo/
└── .github/
    └── skills/
        ├── moodboard-to-html/
        ├── moodboard-to-template/
        └── zine-page/

02 · open chat

Open GitHub Copilot Chat in VS Code.

Click the Copilot icon in the VS Code title bar, or press Ctrl+Alt+I on Windows / Linux (Cmd+Ctrl+I on macOS). The Chat panel opens on the right.

Copilot auto-discovers skills under .github/skills/ — no install command, no build step, no selecting a agent or typing a prompt. If you just dropped the folder into an open workspace, reload the window once (see Step 01) and you're set.

Don't see the Copilot Icon?

If you don't see a Copilot icon, install the GitHub Copilot Chat extension from the Extensions view first.

GitHub Copilot Chat extension card in the VS Code Extensions view, showing the GitHub Copilot Chat name, description, and Verified GitHub publisher badge.

03 · choose a path

Pick the path that matches what you've got.

Path A · with mood board

Drop in images + write a prompt.

Attach 1–5 reference images (inspiration posters, screenshots, swatches). Then ask:

Use the moodboard-to-html skill to
turn these images into an HTML presentation
about how to build cute custom PCB designs.
Mood board collage example — bold editorial, kawaii, and cyber references

Path B · no mood board

Just describe what you want.

No images needed — the skill falls back to a default editorial theme:

Create a presentation
guide to teach what
a Skill is for new users.
Default editorial theme example — warm paper, serif headlines, teal accent

04 · output

Where the file lands.

The skill writes a single self-contained HTML file with CSS inline. Open it in a browser to preview, or commit it to your repo and publish it with GitHub Pages — no build step required.

05 · what it can do

Eight things the skills handle automatically.

You bring the words. The skills handle the look, the rhythm, and the things that are easy to forget.

01 · layout

One canonical page shell.

Every page is masthead → hero with one h1 → body sections with h2 → prev/next pills → footer.

02 · font

Display + body + mono.

One Google Fonts triplet, declared once, loaded on every page.

Display Body copy that reads. mono · 0123

03 · style

Editorial accents.

Italic em emphasis in headlines, pullquotes, lede paragraphs, a consistent masthead-to-footer rhythm.

“The skill is the folder.”

04 · color palette

Bring your own, or use the default.

Bundled cream + navy + mustard, derived from your reference images, or hand-rolled — chosen once in Step 0.

05 · code snippets

Dracula-toned code blocks.

Dark .codeblock with .kw, .str, .cm spans in every page; Dracula duotone code-bg / code-ink tokens in the matching style guide.

---
name: brand-voice
---

06 · interactive code examples

Click to watch it type.

A live typewriter sells the idea better than a screenshot. Click the button; the snippet types itself into the block.

name: brand-voice
rules:
  - Verbs over nouns
  - Cut every adverb
  - No em-dashes in product copy
  - Short sentences only
  - First person plural always

07 · animation

Themed motion, motion-safe.

Inline SVG centerpieces driven by CSS keyframes, wrapped in prefers-reduced-motion guards.

08 · built-in accessibility

Checked before write.

Every page passes the Accessibility Bar before it leaves the skill.

  • Skip link + single h1
  • Landmarks + aria-labels
  • Focus-visible outline
  • AA contrast
  • Honors reduced motion

06 · anatomy

A skill is two things: a YAML header and a Markdown body.

Your agent host reads the header to decide whether to use the skill. The model reads the body to actually do the work.

.github/skills/brand-voice/SKILL.md

---
name: brand-voice
description: Apply our editorial voice —
  short verbs, no adverbs, no em-dashes
  in product copy.
---

# Brand voice

Verbs over nouns. Cut every adverb.
Short sentences only. First person
plural, always.

Same SKILL.md file. Different host. Different model. It still works.

faq · troubleshooting

Common questions.

What is a SKILL.md and how is it different from an agent and a prompt?

The easiest way to keep these straight is to ask who invokes it?

A skill (a SKILL.md file in .github/skills/<name>/) is something the model picks for you. Copilot reads the skill's description from its YAML frontmatter and auto-loads it when your request matches — that's how this repo's three skills get triggered. You can also run one explicitly by typing /skill-name.

A custom agent (an .agent.md file in .github/agents/) is something you pick from the agents dropdown in the Chat view. It's a persistent persona — security reviewer, planner, whatever — with its own instructions, allowed tools, and preferred model, and it stays active until you switch away.

A prompt file (a .prompt.md file in .github/prompts/) is something you type as a slash command. It's a saved one-shot template — never auto-loaded — that runs when you invoke /prompt-name.

Copilot doesn't see my skill — what now?

Double-check the path is exactly .github/skills/<skill-name>/SKILL.md. If it still isn't picked up, reload the VS Code window (Ctrl+Shift+PDeveloper: Reload Window).

Which Copilot model should I use?

Any current GitHub Copilot Chat model works. For Path A, pick a model with vision support — image-aware models give noticeably better mood-board results. Recommended options include Anthropic Opus 4.6 or 4.7, or OpenAI GPT-5.4 or GPT-5.5.

Can I edit the output?

Yes. The output is plain HTML with inline CSS — edit it like any other file, or ask Copilot to iterate on it.