Ai To Figma logo

Ai To Figma

Community
Niall-Young
ai-to-figma

Plan web or iOS mobile interfaces from screenshots, Figma frame references, or a product brief; generate and explicitly approve full-screen visual previews for adaptation and independent design; rebuild approved directions as capture-safe static HTML with official assets; and push editable layers into Figma Design with Figma MCP code-to-canvas. Use when a user asks to design, recreate, reference, or independently plan a screen and deliver it directly to Figma.

Overview

PublisherNiall-Young
RepositoryAItoFigma
Skill nameai-to-figma
Stars
52
Forks
6
Bundled files
34
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.

  • 34 bundled files

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

  • Open source

    Published by Niall-Young on GitHub. Read the source before you install it.

Installation

Install the Ai To Figma 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/Niall-Young/AItoFigma/tree/main/skills/ai-to-figma
  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/Niall-Young/AItoFigma.git /tmp/AItoFigma
mkdir -p .claude/skills
cp -r /tmp/AItoFigma/skills/ai-to-figma .claude/skills/ai-to-figma
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Ai To Figma 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 Ai To Figma 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 Ai To Figma 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.

AI to Figma

Turn a design brief or visual reference into an approved visual direction, rebuild it as validated static HTML, then capture that HTML into an editable Figma Design file.

Core contract

  1. Resolve the brief, reference role, delivery target, fidelity mode, and device before creating the task workspace. Open the delivery-target chooser only for missing target fields. In Codex, insert it with the native ::codex-inline-vis directive; do not substitute a visualize content reference or modal form.
  2. Scaffold design.md with the resolved mode and without creating index.html. For adaptation and independent planning, establish the visual direction, generate a full-screen preview, iterate without a fixed limit, and stop until the user explicitly approves one version.
  3. After explicit approval, persist the selected file under previews/ and run scripts/record-preview-approval.mjs. Reconstruction skips only this full-screen preview gate, not generation of separate final assets.
  4. Resolve the local-font Typography Plan, inventory every non-text final visual in the Visual Asset Plan, classify its complexity as structural, flat, or rich, and show both complete decision tables to the user. Never abbreviate away composition, identity, complexity, method, or status. The approved full-screen preview is a reference, not a final asset.
  5. Resolve every typography and visual-asset row to ready, validate all required gates, then create index.html with scripts/init-html.mjs. Stop on any blocked item.
  6. Rebuild the approved direction with static HTML, inline CSS, separate final assets, and only small amounts of native JavaScript. Mark each planned visual with its data-visual-id; never embed the full-screen preview or introduce React or a remote runtime.
  7. Validate the rendered page and its visual-plan mappings locally before requesting a Figma capture.
  8. Use Figma MCP generate_figma_design as the delivery path and open only the root-scoped captureUrl emitted by scripts/serve-preview.mjs. Do not silently fall back to use_figma, clipboard HTML, SVG import, or a flat screenshot.
  9. Treat every capture ID as single-use. Permit one diagnosed retry with a new capture ID, then stop and report the failure.

Input roles

  • Brief: required product/content intent unless the user explicitly requests a literal reconstruction.
  • Reference: optional screenshot, local image, or node-specific Figma URL used for reconstruction or visual-language extraction.
  • Target: optional prompt input, but required before work begins. Use an explicitly designated Figma Design URL, or resolve the current Figma identity and plan for a new Design file through the delivery-target gate.
  • Device: web or mobile. Infer it only when the prompt or reference makes the class unambiguous.

If one Figma URL could be either a reference or a target, ask which role it has. Never guess.

Interpret “recreate”, “restore”, “1:1”, or equivalent language as reconstruction. Interpret “inspired by”, “reference”, or equivalent language as visual-language adaptation. Otherwise plan the interface from the brief.

Required references

Hard prohibitions

  • Do not crop a UI reference screenshot into final asset files.
  • Do not generate, invent, or approximately redraw an existing brand or product logo.
  • Do not regenerate an identity-critical product, person, package, or mascot and present it as the original.
  • Do not use an unvalidated transparent asset.
  • Do not generate a keyed background or run local background removal for a GPT Image 2 asset; request native transparency and validate the final alpha output.
  • Do not reduce a compositionally dominant hero illustration to generic CSS merely because its silhouette is abstract or geometrically reproducible.
  • Do not implement a rich visual with HTML/CSS, inline SVG, MingCute, or device-native structure. Use a file-backed original or image generation.
  • Do not substitute SVG or CSS when an image-generation row cannot be resolved. Invoke the runtime image-generation capability or mark the row blocked.
  • Do not create index.html before the Visual Asset Plan passes or continue while any visual is planned, resolving, or blocked.
  • Do not create index.html before the Typography Plan passes, use a remote or downloaded font, or leave a requested display font unresolved.
  • Do not create index.html for adaptation or independent planning before an explicitly approved preview is recorded and its hash validates.
  • Do not use the approved full-screen preview in HTML, the Visual Asset Plan, or Figma output; rebuild it as editable structure and separate assets.
  • Do not generate or distort an identity-critical logo, person, product, package, mascot, or character in a direction preview. Leave a neutral placeholder or use an authorized original without alteration.
  • Do not bake required shadows, backgrounds, UI text, or controls into generated subject images.
  • Do not mix MingCute with another generic icon family, use an icon font, retain an icon placeholder, or load icons at page runtime.
  • Do not send unreviewed HTML to Figma.
  • Do not capture body, html, or the browser viewport; capture only [data-figma-capture-root] through the URL emitted by scripts/serve-preview.mjs.
  • Do not claim delivery until the captured node or page has been reread and visually verified.
  • Do not create .ai-to-figma/<slug>/ while the delivery target is unresolved or an account switch is pending.

Bundled helpers

Resolve the directory containing this SKILL.md as AI_TO_FIGMA_SKILL_DIR. Use the bundled scripts from that directory instead of recreating their mechanics; keep generated task files in the user's current project:

bash
AI_TO_FIGMA_SKILL_DIR="/absolute/path/to/ai-to-figma"
npm install --omit=dev --prefix "$AI_TO_FIGMA_SKILL_DIR"
node "$AI_TO_FIGMA_SKILL_DIR/scripts/render-target-chooser.mjs" --input /thread-visuals/target-state.json --output /thread-visuals/figma-target-chooser.html
node "$AI_TO_FIGMA_SKILL_DIR/scripts/scaffold.mjs" --device web --mode adaptation --slug landing-page
node "$AI_TO_FIGMA_SKILL_DIR/scripts/record-preview-approval.mjs" --dir .ai-to-figma/landing-page --file previews/approved-v3.png
node "$AI_TO_FIGMA_SKILL_DIR/scripts/list-local-fonts.mjs" --query "PingFang"
node "$AI_TO_FIGMA_SKILL_DIR/scripts/validate-design.mjs" --file .ai-to-figma/landing-page/design.md
node "$AI_TO_FIGMA_SKILL_DIR/scripts/init-html.mjs" --dir .ai-to-figma/landing-page
node "$AI_TO_FIGMA_SKILL_DIR/scripts/inline-mingcute-icon.mjs" --file .ai-to-figma/landing-page/index.html --name home-1 --style regular --visual-id home-icon
node "$AI_TO_FIGMA_SKILL_DIR/scripts/validate-html.mjs" --file .ai-to-figma/landing-page/index.html --device web --design .ai-to-figma/landing-page/design.md
node "$AI_TO_FIGMA_SKILL_DIR/scripts/serve-preview.mjs" --dir .ai-to-figma/landing-page --port 4173
node "$AI_TO_FIGMA_SKILL_DIR/scripts/serve-preview.mjs" --dir .ai-to-figma/landing-page --port 4173 --inject-file /task-temp/capture-snippet.html --capture-id <capture-id> --capture-endpoint <capture-endpoint> --capture-delay 1000
python3 "$AI_TO_FIGMA_SKILL_DIR/scripts/validate-alpha.py" --input subject-transparent.png --preview-dir alpha-previews
# Legacy compatibility for a user-provided or first-party flat-background asset only:
python3 "$AI_TO_FIGMA_SKILL_DIR/scripts/remove-solid-background.py" --input legacy-source.png --out subject-transparent.png --key black --mode light-on-black
python3 "$AI_TO_FIGMA_SKILL_DIR/scripts/validate-alpha.py" --input subject-transparent.png --preview-dir alpha-previews --source-key black

The generated task directory, approved preview and approval record when required, Figma URL, verification result, asset provenance, and remaining limitations are required final outputs.

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 Ai To Figma AI skill do?

Plan web or iOS mobile interfaces from screenshots, Figma frame references, or a product brief; generate and explicitly approve full-screen visual previews for adaptation and independent design; rebuild approved directions as capture-safe static HTML with official assets; and push editable layers into Figma Design with Figma MCP code-to-canvas. Use when a user asks to design, recreate, reference, or independently plan a screen and deliver it directly to Figma.

Why use Ai To Figma on TypingMind?

Because you install it once and use it with any model. Ai To Figma 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 Ai To Figma in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Niall-Young/AItoFigma/tree/main/skills/ai-to-figma. 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 Ai To Figma?

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 Ai To Figma?

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

Is the Ai To Figma AI skill free?

Yes. It is published on GitHub by Niall-Young 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 👇