Desktop Visual Design logo

Desktop Visual Design

Community
Firstp1ck
desktop-visual-design

Use when building or styling desktop apps, Quickshell QML components, bars, panels, popups, launchers, notifications, or widgets that should fit an existing system theme. Derives reusable palette, interaction, spacing, typography, border, geometry, and motion tokens from the target project and desktop instead of guessing values.

Overview

PublisherFirstp1ck
Repositorypi-coding-agent-forge
Skill namedesktop-visual-design
Stars
93
Forks
22
Bundled files
3
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 3 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by Firstp1ck on GitHub. Read the source before you install it.

Installation

Install the Desktop Visual Design AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/Firstp1ck/pi-coding-agent-forge/tree/main/pi-skill-desktop-visual-design/skills/desktop-visual-design
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/Firstp1ck/pi-coding-agent-forge.git /tmp/pi-coding-agent-forge
mkdir -p .claude/skills
cp -r /tmp/pi-coding-agent-forge/pi-skill-desktop-visual-design/skills/desktop-visual-design .claude/skills/desktop-visual-design
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Desktop Visual Design in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Desktop Visual Design on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Desktop Visual Design is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Desktop Visual Design

Build system-native desktop interfaces from the visual rules already present in the user's environment. Inspect the target first, turn repeated values into semantic tokens, reuse host components, and verify the result across interaction states and theme changes. When the target has no visual contract, use the restrained terminal-first fallback defined here rather than inventing a glossy web-app style.

Installing this Pi package makes the skill available to Pi. It does not modify desktop settings, themes, or application code by itself.

When to use

Trigger for:

  • Quickshell widgets, panels, bars, popups, overlays, launchers, and notifications.
  • Desktop apps built with QML, Qt, GTK, Tauri, Electron, or embedded web views.
  • UI reviews focused on palette, hierarchy, spacing, typography, borders, geometry, focus states, and motion.
  • Requests to make a desktop interface match an existing system theme or first-party surface.

Do not trigger for server-side work, CLI output formatting, compositor configuration unrelated to application appearance, or general public-website design without a desktop-integration requirement.

Principles

  1. Inspect before styling. Existing theme files, token objects, shared controls, and first-party surfaces are the source of truth.
  2. Honor the active color scheme. Standalone apps must read the toolkit, portal, or host-shell light or dark preference before choosing a built-in palette. For Linux Qt apps, prefer an explicit XDG portal result at startup because Qt platform-theme environment can differ between launchers; use Qt style hints as the fallback and live-change input. Automatic mode must update when its authoritative preference changes.
  3. Use semantic roles. Components depend on foreground, background, accent, urgent, muted, and named surface roles instead of literal palette colors.
  4. Define every interaction state. Cover idle, hover, keyboard cursor, focus, selected, pressed, disabled, and urgent states where they apply.
  5. Keep geometry stable. Reserve border and focus-ring space so interaction never shifts layout.
  6. Use one scale. Derive spacing and typography from named tokens and a documented base size.
  7. Retheme through one path. Parse, validate, and publish an atomic theme snapshot. Theme changes must not leave mixed old and new values on screen.
  8. Match existing motion. Honor the target's motion language and reduced-motion settings. When no motion language exists, animate only short color or opacity changes.
  9. Restrain decoration. Without an explicit reference, prefer opaque flat surfaces, one accent family, thin borders, square geometry, and no gradients, glow, blur, hover lift, or decorative shadows.
  10. Compare relationships, not sampled pixels. A screenshot can establish palette relationships, hierarchy, composition, framing, type treatment, landmarks, and state language. It does not override project tokens with guessed colors or justify forcing its apparent color scheme.

Token model

Palette

Start with semantic roles:

RoleUse
foregroundtext, icons, and state washes
backgroundprimary surfaces and scrims
accentselection, focus punctuation, and progress
urgenterrors and critical attention
mutedplaceholders, dividers, and metadata

Add surface-specific roles only when the target already distinguishes them. Do not name component tokens after a hue. The fallback uses a violet-charcoal relationship in dark mode: a charcoal base with a restrained violet cast, slightly separated opaque surfaces, high-contrast foreground, dimmer metadata, and a periwinkle-like structural accent. Its light counterpart uses a softly violet-tinted light base, slightly darker or brighter opaque surfaces as needed for separation, dark foreground, dimmer metadata, and a deeper related structural accent. These are relationships, not literal colors. Selection, focus, and progress share the accent family; semantic success gets a distinct success role only when the interface needs it, and urgent remains reserved for errors. Do not introduce multi-hue decorative palettes, gradients, glow, or translucent glass unless the target already uses them.

Interaction states

If the target project has no state model, use this compact reference profile:

StateFill alphaBorder alphaBorder width
normal0.000.351px
hover or keyboard cursor0.060.601px
focus0.061.001px
selected0.160.721px
pressed0.220.881px
text selection0.30none0px

Apply state precedence in this order: pressed, focus, hover or cursor, selected, active, idle. Use visible focus even when hover and focus share a fill.

Spacing and typography

At a 12px base font, the fallback spacing scale is:

xxs 2, xs 4, sm 6, md 8, lg 10, xl 12, xxl 16, xxxl 20, huge 24

Fallback type multipliers are caption 0.833, body-small 0.917, body 1.0, subtitle 1.083, title 1.167, heading 1.333, display 2.0, and display-large 2.333. Prefer the desktop's monospace UI alias across controls, labels, and data. Use display sizes selectively for the primary title, value, or empty-state landmark. Use uppercase labels with restrained 0.08em to 0.12em tracking for short section headings, not for paragraphs or long actions.

The target project's own values always win. See references/design-tokens.md for the complete fallback profile.

Geometry and motion

Read geometry from project tokens or compositor settings when available. On Hyprland, hyprctl -j getoption decoration:rounding and general:gaps_out can inform radius and edge spacing. Validate command output and keep the last valid value on failure.

Without a geometry source, use square panels and controls with a 0px radius, allowing up to 2px only when needed to avoid harsh raster clipping. Use opaque surfaces, 1px borders, and simple separators. Let an outer frame, major module borders, and a small number of separators carry hierarchy; do not wrap every row in a card. Keep status and selection treatments rectangular. Do not use pill geometry unless shape itself communicates a compact status or toggle.

Without an existing motion language, use 90 to 120ms color or opacity transitions and no positional hover movement. Keep functional progress animation, such as a spinner, constant and subtle. Reduced-motion mode removes nonessential animation.

Workflow

  1. Confirm the target. Identify the toolkit, host shell or app, supported interaction methods, scale factors, supported theme modes, and the current system color-scheme preference. Completion: the runtime, active preference, and visual comparison target are explicit.
  2. Inventory the visual contract. Find palette sources, color-scheme signals, theme objects, token modules, shared controls, font settings, geometry sources, motion constants, and first-party reference surfaces. Completion: each planned value has an owner or is marked as a fallback.
  3. Analyze supplied screenshots. Record observations separately from implementation choices. Compare palette relationships; primary, secondary, and display hierarchy; density and composition; outer and module framing; typography scale and tracking; sparse functional landmark iconography; and idle, hover, focus, selected, pressed, disabled, success, and urgent state language. Also record what the image cannot establish, including exact colors, unseen states, automatic-mode behavior, and the opposite color scheme. Completion: each observation is mapped to a project-owned token, a semantic fallback role, or an explicit unknown.
  4. Choose the integration path. Inside a host project, reuse its theme and control modules. For a standalone app, define source precedence for automatic mode and an explicit app-owned input only for user overrides. A Linux Qt app should prefer a valid XDG portal light or dark result, then fall back to Qt style hints. Read the matching reference file.
  5. Create semantic tokens. Map source values into complete light and dark palette, surface, state, spacing, typography, border, geometry, and motion roles. Keep literal fallback colors in the theme owner. Completion: UI components own no palette literals.
  6. Implement complete states. Add pointer, keyboard, focus, selected, pressed, disabled, success, and urgent behavior as needed. Completion: keyboard use is visible and layout remains stable.
  7. Wire theme changes. Subscribe to the toolkit, portal, host-shell, or validated file signal that owns the preference. Publish all changed tokens together and retain the last valid snapshot on input failure.
  8. Compare the result. Place current captures beside the references at representative scale. Judge whether the same relationships hold across palette, hierarchy, density/composition, framing, type scale/tracking, landmarks, and states—not whether pixels match. Repeat in both color-scheme branches and document intentional differences.
  9. Verify. Force light and dark branches, return to automatic mode, change the live system preference, compare first-party surfaces, test scaling and input modes, scan components for palette literals, and run the project's checks.

Invocation design

  • Strong signals include desktop UI, Quickshell, QML, panel, popup, launcher, notification, native-looking, system theme, live retheme, and visual consistency.
  • Branches are host-project component, standalone desktop app, portable web-view integration, theme-system design, and visual review.
  • Do not route compositor keybinds, backend work, package maintenance, or unrelated web branding here.

References

  • references/design-tokens.md defines the fallback token profile and explains when project values take priority.
  • references/quickshell-plugin-styling.md covers components that run inside an existing Quickshell configuration.
  • references/portable-theme-loading.md covers explicit theme inputs, atomic reloads, Hyprland geometry, QML, and web-view mappings.

Verification

  • Search changed UI components for literal palette colors. Keep fallback literals only in the theme owner and justify any exception.
  • Test pointer and keyboard focus, selected, pressed, disabled, and urgent states.
  • Force both light and dark branches, then run a separate automatic-mode acceptance check with no test override and confirm it matches the desktop preference.
  • Test no-preference, malformed, and unavailable source results plus any disagreement precedence between toolkit and portal values.
  • Change the live desktop preference and confirm every semantic color tied to the live source updates together without restarting the app.
  • Test supported scale factors, font changes, and bar or panel edges.
  • Compare current screenshots beside the supplied reference and existing first-party surfaces. Record pass, difference, or not observable for palette relationships; hierarchy; density and composition; outer and module framing; typography scale and short-label tracking; sparse functional landmark icons; and state language. Verify meaning and relative emphasis rather than sampled colors or exact coordinates.
  • When the fallback is active, also check for flat opaque surfaces, a coherent light counterpart to the dark relationship, monospace UI text, selective display scale, square geometry, stable 1px focus geometry, restrained accent use, rectangular states, and the absence of decorative glow or hover lift. Green or another success color must remain semantic punctuation rather than general selection decoration. For Quickshell windows, confirm the visual root belongs to contentItem and that intended opaque surfaces do not reveal the windows behind them.
  • Run the target project's documented formatting, type, and UI checks.

Safety and failure modes

  • Keep inspection read-only until the user authorizes UI changes.
  • Do not edit desktop settings or theme sources when the request only covers an app or component.
  • Treat theme files and environment values as untrusted input. Validate colors, numbers, paths, and width lists.
  • Use argument arrays for external commands. Never build shell commands from theme values.
  • A missing or unknown system color-scheme preference uses the toolkit palette or a documented neutral fallback. It must not silently force light mode.
  • Missing theme data uses a documented fallback and reports that fallback once.
  • Invalid reloads retain the last valid theme instead of partially applying values.
  • Make targeted edits to files containing private-use or Nerd Font glyphs.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Desktop Visual Design AI skill do?

Use when building or styling desktop apps, Quickshell QML components, bars, panels, popups, launchers, notifications, or widgets that should fit an existing system theme. Derives reusable palette, interaction, spacing, typography, border, geometry, and motion tokens from the target project and desktop instead of guessing values.

Why use Desktop Visual Design on TypingMind?

Because you install it once and use it with any model. Desktop Visual Design is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Desktop Visual Design in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Firstp1ck/pi-coding-agent-forge/tree/main/pi-skill-desktop-visual-design/skills/desktop-visual-design. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Desktop Visual Design?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Desktop Visual Design?

As many as you like. As long as a model supports skills, you can use Desktop Visual Design with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Desktop Visual Design AI skill free?

Yes. It is published on GitHub by Firstp1ck under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇