Stitch::Generate Design logo

Stitch::Generate Design

OrganizationPopular
google-labs-code
stitch::generate-design

Generate new screens from text prompts or images, edit existing screens with prompts and design system tokens, and generate design variants using Stitch MCP. Includes prompt enhancement pipeline, design mappings, professional UI/UX terminology, design tokens and theme system capabilities.

Overview

Publishergoogle-labs-code
Repositorystitch-skills
Skill namestitch::generate-design
Stars
8.3K
Forks
1.1K
Bundled files
3
LicenseApache-2.0
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 google-labs-code on GitHub. Read the source before you install it.

Installation

Install the Stitch::Generate 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.
  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/google-labs-code/stitch-skills.git /tmp/stitch-skills
mkdir -p .claude/skills
cp -r /tmp/stitch-skills/plugins/stitch-design/skills/generate-design .claude/skills/google-labs-code-stitch-generate-design
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Stitch::Generate 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 Stitch::Generate 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 Stitch::Generate 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.

Generate Design

Create new design screens from text descriptions, images, or mockups, edit existing screens with prompts and design system tokens, and generate design variants using Stitch MCP.

[!NOTE] Refer to your system prompt for instruction on handling MCP tool prefixes for all tools mentioned in this skill (e.g., list_projects, generate_screen_from_text, edit_screens).

🎨 Prompt Enhancement Pipeline

Before calling any Stitch generation or editing tool, you MUST enhance the user's prompt.

1. Analyze Context

  • Project: Use list_projects to find the correct projectId. If no suitable project exists, create one using create_project.
  • Design System: Check if a design system exists for the project via list_design_systems. If one exists, design tokens (colors, fonts, roundness) are already applied at the project level — do NOT include any color, font, or theme instructions in the generation prompt. If none exists, delegate to the manage-design-system skill first before generating screens.

2. Refine UI/UX Terminology

Consult Design Mappings to replace vague terms.

  • Vague: "Make a nice header"
  • Professional: "Sticky navigation bar with glassmorphism effect and centered logo"

Use Prompting Keywords for component names, adjective palettes, color roles, and shape descriptions.

3. Structure the Final Prompt

Format the enhanced prompt for Stitch. Focus exclusively on layout, content, and structure — never include colors, fonts, or theme instructions (these are handled by the manage-design-system skill at the project level).

For new screens, use this template:

markdown
[Overall purpose and user intent of the page]

**PLATFORM:** [Web/Mobile], [Desktop/Mobile]-first

**PAGE STRUCTURE:**
1. **Header:** [Description of navigation and branding]
2. **Hero Section:** [Headline, subtext, and primary CTA]
3. **Primary Content Area:** [Detailed component breakdown]
4. **Footer:** [Links and copyright information]

For edits, be specific about what to change:

  • Location: "Change the [primary button] in the [hero section]..."
  • Visuals: "...to a darker blue (#004080) and add a subtle shadow."
  • Structure: "Add a secondary button next to the primary one with the text 'Learn More'."

[!CAUTION] Do NOT include hex codes, font names, color palettes, roundness values, or any design system tokens in a generation prompt. These are applied at the project level by the manage-design-system skill and will conflict if duplicated. (For edit prompts, hex codes are acceptable for precise color adjustments.)

4. Present AI Insights

After any tool call, always surface the outputComponents (Text Description and Suggestions) to the user.

See examples/enhanced-prompt.md for a full before/after prompt enhancement example.


Steps

Determine the Mode

Decide which flow to use based on the user's request:

  • User wants to create from a text description → Generate from Text flow
  • User provides an image, screenshot, or mockup → Generate from Image flow
  • User wants to modify an existing screen → Edit flow
  • User wants layout/color/content variations → Generate Variants flow

Generate from Text Flow (New Screen)

1. Enhance the User Prompt

Apply the Prompt Enhancement Pipeline above.

2. Identify the Project

Use list_projects to find the correct projectId if it is not already known.

3. Generate the Screen

Call the generate_screen_from_text tool with the enhanced prompt and the designSystem ID (if found in Step 1).

json
{
  "projectId": "...",
  "prompt": "[Your Enhanced Prompt]",
  "designSystem": "assets/...", // Optional: Pass if found in Step 1
  "deviceType": "DESKTOP"  // Options: MOBILE, DESKTOP, TABLET
}
4. Present AI Feedback

Always show the text description and suggestions from outputComponents to the user.

5. Download Design Assets

After generation, download the HTML and screenshot urls from outputComponents to the .stitch/designs directory.

  • Naming: Use the screen ID or a descriptive slug for the filename.
  • Tools: Use curl -o via run_command or similar.
  • Directory: Ensure .stitch/designs exists.
6. Review and Refine
  • If the result is not exactly as expected, continue with the Edit flow to make targeted adjustments.
  • Do NOT re-generate from scratch unless the fundamental layout is wrong.

Generate from Image Flow (Image/Mockup → Design)

Use this flow when the user provides an image, screenshot, or design mockup to recreate in Stitch.

1. Identify the Project

Use list_projects to find the correct projectId. If no suitable project exists, create one using create_project.

2. Upload the Image

Delegate to the upload-to-stitch skill to upload the image to the project. This creates a new screen with the image as its content.

3. Refine with Edit

Once uploaded, use list_screens to find the newly created screenId, then call edit_screens with a descriptive prompt to refine the design:

json
{
  "projectId": "...",
  "selectedScreenIds": ["<uploaded-screen-id>"],
  "prompt": "[Describe what to adjust, enhance, or recreate from this mockup]"
}

[!TIP] For best results, describe the intent behind the image rather than just saying "make it look like this". For example: "This is a dashboard mockup — recreate it with a proper data table, sidebar navigation, and chart widgets."

4. Present AI Feedback

Always show the text description and suggestions from outputComponents to the user.

5. Download Design Assets

Download the HTML and screenshot urls from outputComponents to the .stitch/designs directory.

  • Naming: Use the screen ID or a descriptive slug for the filename.
  • Tools: Use curl -o via run_command or similar.
  • Directory: Ensure .stitch/designs exists.

Edit Flow (Modify Existing Screen)

1. Identify the Screen

Use list_screens or get_screen to find the correct projectId and screenId.

2. Formulate the Edit Prompt

Apply the Prompt Enhancement Pipeline, focusing on specificity:

  • Location: "Change the color of the [primary button] in the [hero section]..."
  • Visuals: "...to a darker blue (#004080) and add a subtle shadow."
  • Structure: "Add a secondary button next to the primary one with the text 'Learn More'."
3. Apply the Edit

Call the edit_screens tool.

json
{
  "projectId": "...",
  "selectedScreenIds": ["..."],
  "prompt": "[Your targeted edit prompt]"
}
4. Present AI Feedback

Always show the text description and suggestions from outputComponents to the user.

5. Download Design Assets

After editing, download the updated HTML and screenshot urls from outputComponents to the .stitch/designs directory, overwriting previous versions to ensure the local files reflect the latest edits.

  • Naming: Use the screen ID or a descriptive slug for the filename.
  • Tools: Use curl -o via run_command or similar.
  • Directory: Ensure .stitch/designs exists.
6. Update Project Metadata

After downloading assets, update .stitch/metadata.json to reflect any changes (e.g., updated screen titles or new screen IDs from the edit). The metadata file tracks all screens, their device types, and design system info. See the manage-design-system skill's examples/metadata.json for the format.

7. Verify and Repeat
  • Check the output screen to see if the changes were applied correctly.
  • If more polish is needed, repeat the edit flow with a new specific prompt.

Generate Variants Flow (Explore Variations)

Use this flow when the user wants to explore alternative layouts, color schemes, or content variations of an existing screen.

1. Identify the Screen

Use list_screens or get_screen to find the correct projectId and screenId.

2. Configure Variant Options

Call the generate_variants tool with the appropriate options:

json
{
  "projectId": "...",
  "selectedScreenIds": ["..."],
  "prompt": "[Describe the direction for variants]",
  "variantOptions": {
    "variantCount": 3,
    "creativeRange": "EXPLORE",
    "aspects": ["LAYOUT", "COLOR_SCHEME"]
  }
}

Variant Options:

  • variantCount: 1–5 variants (default: 3)
  • creativeRange: REFINE (subtle), EXPLORE (balanced), or REIMAGINE (radical)
  • aspects: Focus on specific dimensions — LAYOUT, COLOR_SCHEME, IMAGES, TEXT_FONT, TEXT_CONTENT, or leave empty for all
3. Present AI Feedback

Always show the text description and suggestions from outputComponents to the user.

4. Download Design Assets

Download the variant HTML and screenshot urls from outputComponents to the .stitch/designs directory.

  • Naming: Use the screen ID or a descriptive slug for the filename.
  • Tools: Use curl -o via run_command or similar.
  • Directory: Ensure .stitch/designs exists.

💡 Tips

  • Be structural: Break the page down into header, hero, features, and footer in your prompt.
  • Content first: Describe what each section contains (text, images, CTAs) rather than how it looks.
  • Iterative Polish: Prefer editing for targeted adjustments over full re-generation.
  • No theme leakage: Never put hex codes, font names, or color roles in a generation prompt — the design system handles all visual styling.
  • Specify interactions: Mention hover states, animations, and click behavior rather than visual styling.
  • Keep edits focused: One edit at a time is often better than a long list of changes.
  • Reference components: Use professional terms like "navigation bar", "hero section", "footer", "card grid".
  • Precise colors in edits: Use hex codes for exact color matching when editing existing screens.

📚 References

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 Stitch::Generate Design AI skill do?

Generate new screens from text prompts or images, edit existing screens with prompts and design system tokens, and generate design variants using Stitch MCP. Includes prompt enhancement pipeline, design mappings, professional UI/UX terminology, design tokens and theme system capabilities.

Why use Stitch::Generate Design on TypingMind?

Because you install it once and use it with any model. Stitch::Generate 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 Stitch::Generate Design in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/google-labs-code/stitch-skills/tree/main/plugins/stitch-design/skills/generate-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 Stitch::Generate 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 Stitch::Generate Design?

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

Is the Stitch::Generate Design AI skill free?

Yes. It is published on GitHub by google-labs-code under the Apache-2.0 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 👇