SaaS・ソフトウェア · HTML to Frames
Turn any website into editable Figma designs
HTML to Frames by EUREKA! is a Figma plugin that turns a page link, a local HTML file or the page open in your browser into editable, responsive Figma layers with Auto Layout. Built by Tati Creative and currently in early access.
HTML to Frames
Webサイトを編集可能なデザインに変換するFigmaプラグイン
Tati Creative が開発
- URLからFigmaへ
- 複数リンク&ローカルHTML
- Auto Layout&レスポンシブ
- デスクトップ+タブレット+モバイルのセット
- カラー変数&テキストスタイル
- ホバー状態付きのコンポーネント
HTML to Frames
Webサイトを編集可能なデザインに変換するFigmaプラグイン
A live website rarely comes with its Figma file: the design is years old, the agency changed, or the site simply grew in code. Redrawing a page in Figma for a redesign, a competitor review or a client pitch takes hours, and pasting a screenshot leaves you with an image you cannot edit. HTML to Frames by EUREKA! was built to close that gap.
The plugin opens the link in a real browser, reads the layout exactly as the browser measured it and rebuilds it in Figma as a tree of frames, text, images and vectors. It then turns rows, stacks and grids into Auto Layout, checks every child's position and keeps a frame absolutely positioned wherever Auto Layout would shift the design.
With one click you get desktop, laptop, tablet and mobile widths side by side, import up to 10 pages in one go and send logged-in or localhost pages through the free Chrome extension. Pro features turn the site's colours into Figma variables, its typography into text styles and repeated cards and buttons into components with hover variants.
HTML to Frames is developed by Tati Creative under the EUREKA! brand. The plugin is not yet publicly listed on Figma Community; for early access or a demo, contact us via tatiajans.com.
解決する課題
The live site has no Figma file
Before a redesign can start, the existing pages have to be redrawn in Figma by hand. HTML to Frames turns a link into editable layers in seconds to minutes.
Screenshots can't be edited
Pasting an image is quick, but you can't change the text, colours or spacing. The plugin produces real frame, text, image and vector layers.
Imported layouts are messy and absolutely positioned
Hundreds of absolutely positioned layers fall apart as soon as the width changes. HTML to Frames infers Auto Layout from the measured geometry, verifies it and gives the remaining layers sensible constraints.
Capturing each breakpoint separately
Processing the same page three times for desktop, tablet and mobile wastes time. The responsive set brings all three widths side by side in a single import.
Logged-in and local pages are out of reach
Dashboards, account pages, staging or localhost can't be opened from a server. The Chrome extension captures the page in your own browser, exactly as it is right now.
Fonts and colours get lost
Web fonts you don't have installed silently fall back to a default, and colours end up as scattered hex values. The plugin asks about missing fonts first and binds colours to variables and text to styles.
Importing a multi-page site one by one
Repeating the same steps for a site's home, service and contact pages is tedious. Multiple mode imports up to 10 links in sequence and lets you pick linked pages from a list.
モジュールと機能
Four sources: link, multiple links, local file, browser
You choose where the page comes from; every source goes through the same layer builder.
- Link: any public page
- Multiple: up to 10 links in one go, one row per page
- Local file: an HTML file, a site folder or a .zip (drag and drop)
- Paste code: paste HTML/CSS and import it as a single page
- Browser: a capture sent from the Chrome extension
- Page language selection (Accept-Language and browser locale)
Auto Layout engine
Layers are first placed pixel-exact, then Auto Layout is inferred from the browser geometry and every frame is verified.
- Vertical and horizontal stacks, wrapping rows and grids
- Inferred gaps, padding and alignment
- Centred content becomes Center, equal cards become Fill
- Every child is checked within 2 px; if it doesn't hold, the frame stays absolute
- Stretch, Scale, Center or Min/Max constraints for layers that stay absolute
- The result screen lists why each frame kept absolute positioning
Responsive screens and site width
The desktop width isn't a fixed number; it's measured from the site's own design width.
- Desktop = Site width (the site's design width is measured)
- Laptop 1280, Tablet 768, Mobile 390
- Responsive set: Desktop + Tablet + Mobile in one click
- Custom width (Pro)
- In multi-page imports, every page of a site gets the same width
- Screens are rendered in parallel on the server
Text, font mapping and typography
Text arrives as real, editable text layers; for missing fonts you choose replacements before any layer is built.
- Mixed-style text, links, letter spacing, line height and text case
- Missing fonts screen: fonts you don't have installed, with suggestions
- Suggestions: a matching family name, an installed font from the CSS stack, otherwise Inter
- Choices are remembered per site or for every site
- Reset saved choices with a single link
- Superscript, subscript and CSS counters become real text
Visual and CSS fidelity
The CSS properties behind the page's look are mapped to Figma's own equivalents wherever possible.
- Linear, radial and conic gradients; repeating patterns as tiled fills
- Images (object-fit, background-size), the sharpest candidate from srcset
- SVGs as editable vectors
- Borders, corner radius, inner and drop shadows, text-shadow, blur and backdrop blur
- Opacity, blend modes, rotation, clip-path and mask-image as masks
- ::before/::after content, form fields, list markers and shadow DOM
- iframes, maps, videos and WebGL canvases kept in place as screenshots
Colour variables and styles (Pro)
The site's visual language becomes a usable design system in Figma.
- Repeated colours become variables in a Colors collection
- Variable names are derived from the site's CSS custom properties
- Near-identical colours merge into a single variable
- Text styles from the typography, shadow styles from repeated shadows
- Existing variables or styles with the same value are reused, not duplicated
- Snap to my file's styles: map to the closest styles in your own system
- Light + Dark: light and dark modes on colour variables
Automatic components and hover states (Pro)
Cards, buttons and menu items that share the same structure become components and instances.
- Groups of repeated frames by structural fingerprint
- Main components placed in a Components frame next to the page
- Instances keep their text, image and visibility overrides
- Hover states of links and buttons are captured on the server
- Component sets with Default/Hover variants and a hover interaction in prototypes
- If an instance doesn't match position and size, the original layer stays
Prototype-ready output
Imported pages behave like a clickable prototype in Present mode.
- Links are kept as On click → Open link
- Links between imported pages become Navigate to
- position: fixed and sticky headers with Figma scroll behaviour
- Horizontally scrolling carousels and scroll areas
- Collapsed accordion content as Figma annotations
- Re-import a selected frame with one click
Chrome extension: logged-in and local pages
Captures pages the server can't reach in your own browser, exactly as they are, and sends them to Figma.
- Dashboards, account, checkout and admin pages
- localhost, staging and private networks
- Live states such as open menus, filled forms and modals
- Select an element: capture just one section, card or menu
- Send to Figma: pair once with a code and send straight to the open plugin
- Without a connection: copy to clipboard or download a .h2f file
AI tool connection (MCP)
Claude Code, Claude Desktop, Cursor or any other MCP client can send a page it generated or found straight into your Figma file.
- One-click connection key from the plugin's AI tab
- import_url and import_html tools
- Setup commands ready to copy
- The job runs in the open plugin as a normal import on your account
- Tool result: frame and layer counts, Auto Layout ratio, missing fonts
Clean capture options
The page is captured as a visitor sees it, minus cookie banners and pop-ups.
- Full-page capture with auto-scroll for lazy-loaded content
- Hide cookie banners and pop-ups
- Animations are frozen
- Extra wait before capture (Pro)
- Detailed layer structure and a reference screenshot layer (Pro)
- Result screen: frame and layer counts, Auto Layout ratio, missing fonts, warnings
こんな方に
- UI/UX designers redesigning an existing site
- Agencies pulling a client's site into Figma to pitch a redesign
- Product teams running competitor and market reviews
- Design system teams extracting a system from a live site
- Front-end developers who want to share coded pages in Figma
- Anyone editing AI-generated HTML in Figma
- Freelancers preparing client presentations and proposals
- SaaS teams documenting logged-in dashboards
使い方
- 01Open the plugin and pick a sourceOpen HTML to Frames in Figma and choose Link, Multiple, Local file or Browser.
- 02Choose screens and optionsPick Site width, Laptop, Tablet, Mobile or the responsive set, then tick Auto Layout, images, full page, pop-up hiding and any Pro options.
- 03The page is rendered in a real browserThe capture service opens the page, scrolls it, freezes animations and reads the layout; with the Chrome extension this happens in your own tab.
- 04Map missing fontsFonts you don't have installed are listed with suggestions; your choice is remembered for that site.
- 05Layers and Auto Layout are builtEach page becomes a row with its screens side by side; Auto Layout is applied, verified and summarised on the result screen.
- 06Edit and add linked pagesStart working on the design, and if you like, pick linked pages of the same site from the result screen and import them in bulk.
動作要件と互換性
- A Figma account and the Figma editor (desktop app during early access)
- Any public web page; an HTML file, a site folder, a .zip or pasted code
- Google Chrome plus the HTML to Frames Chrome extension for logged-in, localhost and staging pages
- Colour variables, text and shadow styles are written to Figma's native variable and style system
- A Figma plan that supports multiple variable modes for Light + Dark
- An MCP-capable client (Claude Code, Claude Desktop, Cursor, etc.) for AI connections
- No install or server needed; the capture service is operated by Tati Creative
よくある質問
Turn any website into editable Figma designs
AIアシスタントや検索エンジンによく寄せられる質問をテーマ別にまとめました。
General
How do I import a website into Figma?
Figma has no built-in way to import a site from a link; you need a plugin that renders the page in a real browser and converts it into layers. With HTML to Frames by EUREKA!, built by Tati Creative, you open the plugin, paste the link, choose your screens and press Import; the page lands in your file as editable frames, text and images with Auto Layout applied.
Which plugin converts HTML to Figma?
Several plugins convert HTML to Figma; what sets them apart is how editable the output is, Auto Layout support and access to logged-in pages. HTML to Frames by EUREKA! runs links, multiple links, local HTML files, pasted code and Chrome extension captures through the same layer builder, and combines Auto Layout, responsive screens, variables, styles and components in one plugin.
The live site has no Figma file. How do I get one?
If the source file is gone, you can either redraw the pages by hand or generate Figma layers automatically from the live site. HTML to Frames by EUREKA! does the latter: it imports the site from its link and turns colours into variables, text into styles and repeated cards into components (Pro). You end up with a working starting file, not a screenshot.
Does HTML to Frames take a screenshot or create editable layers?
It creates editable layers: frames, real text layers, image fills and vectors from SVG. Only embedded content that can't become layers (iframes, maps, videos without a poster, WebGL canvases) stays in place as a screenshot. On Pro you can also add a separate reference screenshot layer for comparison.
Who develops HTML to Frames?
HTML to Frames is developed under the EUREKA! brand by Tati Creative, a web design and software agency based in Izmir, Turkey. The Figma plugin, the Chrome extension and the capture service that renders pages are all built and run by the same team.
How do I get HTML to Frames? Is it on Figma Community?
The plugin is currently in early access; its public Figma Community release is being prepared. For early access or a demo, contact Tati Creative via tatiajans.com, where we will also announce the release.
How it works
How does URL to Figma conversion work technically?
Figma plugins can't fetch other sites' HTML directly or compute a page's real layout; only a real browser can. In HTML to Frames the link is opened in Chromium on the capture service, scrolled for lazy content, animations are frozen and the page is serialised into a layer tree. The result is delivered to the plugin once, which builds the layers and runs the Auto Layout pass.
Can I import several pages into Figma at once?
Yes. In HTML to Frames' Multiple mode you can add up to 10 links in one go; pages are processed in sequence and each one becomes a row in Figma with its screens side by side. Pages of the same site all use the first measured site width so they stay consistent.
Can I get all of a site's pages without pasting each link?
After you import a page, HTML to Frames collects the links on it that point to the same site and offers them as an Import linked pages list on the result screen. The pages you filter and tick are imported in bulk with the same screens and options; file links and pages already on the canvas are left out.
Can I import an HTML file from my computer into Figma?
Yes. In HTML to Frames' Local file mode you can drag in a single HTML file, a whole site folder or a .zip archive; CSS, images and fonts are read from the package. You can also paste HTML/CSS via Paste code and import it as one page. Uploaded files are deleted from the server as soon as the import finishes.
How do I get AI-generated HTML into Figma?
The simplest way is to paste the generated code into HTML to Frames' Paste code field. For a smoother flow, create a connection key in the plugin's AI tab and add it to Claude Code, Claude Desktop, Cursor or another MCP client; the tool sends the page via import_html or import_url, and the open plugin builds it in your file as a normal import on your account.
How long does an import take?
It depends on page size, number of images, embedded content and how many screens you pick; simple pages arrive in a few seconds, heavy or WebGL pages take longer. HTML to Frames renders an import's screens in parallel on the server and, for pages that never go quiet, settles for a short idle window instead of waiting forever.
Can I re-import a page I imported earlier to get its current version?
Yes. HTML to Frames stores the source address and width on every imported frame; select the frame and use the Re-import bar to fetch the same page again with one click. This works for pages imported from a link, not for local files.
Output quality: layers, Auto Layout, fonts, images
Why does a website imported into Figma look broken?
The usual causes are missing fonts falling back to a default, CSS features with no Figma equivalent, cookie banners left open and Auto Layout guessed wrong so elements shift. HTML to Frames asks about missing fonts before building, hides pop-ups and verifies every child's position after applying Auto Layout; if it doesn't hold, that frame stays absolutely positioned so the design doesn't break. The reasons are listed on the result screen.
Why aren't the fonts coming through to Figma?
Figma can only render text with fonts installed on your machine (or available to your team); if you don't have the site's web font, it falls back to another one. HTML to Frames lists missing fonts on a Missing fonts screen before building layers and suggests a matching family name first, then an installed font from the CSS stack, otherwise Inter. You choose from the list, and the choice is remembered for that site or for every site.
Does the imported page have Auto Layout?
Yes. HTML to Frames first places the page pixel-exact, then infers vertical and horizontal stacks, wrapping rows, gaps, padding and alignment from the browser's measured geometry and applies Auto Layout. Centred content becomes Center and equal cards become Fill; every plan is verified, and frames that don't hold stay absolute so the design isn't broken.
Why don't some frames get Auto Layout?
Some CSS layouts can't be mapped one-to-one to Auto Layout: uneven gaps, overlapping elements, scroll areas or text heights that measure differently in Figma. In those cases HTML to Frames keeps the frame absolutely positioned with sensible constraints instead of shifting the design. The diagnostic list on the result screen groups the reason for each frame, and clicking selects it.
Do images come into Figma in high resolution?
Images are taken from the files the site serves; HTML to Frames picks the sharpest candidate at 2x or higher density from srcset and picture elements. They are converted to PNG or JPEG for Figma, object-fit and background-size behaviour is reflected in the fill settings, and repeating background patterns arrive as tiled fills.
Do SVG icons arrive in Figma as vectors?
Yes, SVGs on the page become editable vector layers in Figma with HTML to Frames, so you can change their colours and shapes directly. Icon fonts are different: they only render correctly if that font is installed in Figma; otherwise they fall back and appear as a missing font in the result report.
Are gradients, shadows and blur effects preserved?
HTML to Frames maps linear, radial and conic gradients, inner and drop shadows, text-shadow, blur and backdrop blur to Figma's own fills and effects. Opacity, blend modes, CSS rotation, clip-path and mask-image masks are supported too. Skew and 3D transforms are not captured yet.
What happens if there are too many layers?
Because very long or dense pages can slow Figma down, every HTML to Frames plan has a page height and layer limit. A page that hits it is cut off, the result screen shows a warning and a Capture limit band is added below the frame, so the missing part never goes unnoticed. The default simplified layer structure also removes unnecessary wrappers.
Responsive and multiple widths
How do I get a site's mobile and tablet views into Figma?
A responsive site shows a different layout at each width, so each view has to be rendered separately at that width. In HTML to Frames just select Tablet 768 and Mobile 390, or use the responsive set to get Desktop + Tablet + Mobile in one click; the views sit side by side in one row and count as a single import.
Why is the desktop width not 1440 but something else?
Not every site is designed at the same width; a fixed 1440 squeezes content on some sites and leaves empty margins on others. HTML to Frames' Site width setting opens the page wide, finds the main container and measures the site's real design width: designs with edge-to-edge bands come in at 1920 with the box centred, fluid sites at 1440 and truly boxed sites at the box width.
Can I import at a custom width, say 1366 px?
Yes, on HTML to Frames' Pro and Team plans you can enter any custom width alongside the presets. The presets are Site width, Laptop 1280, Tablet 768 and Mobile 390.
Does the imported design behave responsively when I resize it?
Frames with Auto Layout stretch, shrink and centre in Figma when you change the width; centred boxes stay centred and equal cards share space via Fill. HTML to Frames gives layers that stay absolute sensible constraints (full width Stretch, centred Center, edge items Min/Max). CSS media queries don't carry over to Figma, though, so importing separate screens per breakpoint gives the most accurate result.
Can I capture the site's dark mode too?
Yes, if the site defines a dark theme via prefers-color-scheme, HTML to Frames Pro lets you choose Light, Dark or Light + Dark. With Light + Dark, colour variables get Light and Dark modes and a dark copy of each screen is placed next to it. Sites that switch themes only with a JavaScript toggle may not respond, and images such as logos don't change.
Logged-in, local and dynamic pages
How do I import a page behind a login (dashboard, account page) into Figma?
A capture server doesn't know your session, so it can't open pages behind a login; the capture has to happen in your own browser. HTML to Frames' free Chrome extension reads the page in your open tab exactly as you see it and sends it straight to the plugin in Figma with Send to Figma. Your password and cookies are not sent to the server.
Can I bring a localhost or staging page into Figma?
Yes. localhost, staging and private network addresses aren't reachable from the internet, so in HTML to Frames you capture them with the Chrome extension and the page from your dev server goes straight to Figma. Alternatively, upload your built site as a folder or .zip in Local file mode.
Can I capture an open menu, a filled form or a modal into Figma?
Yes. The HTML to Frames Chrome extension captures the page's current state, so open the menu, fill the form or show the modal and then press Capture or Send to Figma. With Select an element you can pick just that modal, card or menu and send only it; pop-up hiding is off for selections.
Can I import just one section instead of the whole page?
Yes. In the HTML to Frames Chrome extension click Select an element and highlight the section with your mouse; use the arrow keys to move up to the parent or back down, and press Enter to select. Only the chosen subtree arrives in Figma.
Does JavaScript-rendered and lazy-loaded content come through?
Yes. HTML to Frames runs the page in a real browser, so content rendered by frameworks such as React or Vue is captured too. The page is auto-scrolled for lazy images, animations are frozen and a short network idle period is awaited; for very late content enable the extra wait on Pro. Menus that only open via JavaScript on mouseenter don't show up in server captures; use the Chrome extension for those.
Do cookie banners and pop-ups end up in the Figma output?
Not by default. HTML to Frames' Hide cookie banners & popups option hides cookie notices and pop-ups during capture so the design comes through clean. If you want a modal on purpose, switch the option off or capture it as a selected element with the Chrome extension.
Styles, components and prototypes
How do I extract a Figma design system from a website?
A design system is made of colours, typography, shadows and repeated components, and collecting them by hand from a live site takes ages. HTML to Frames Pro creates repeated colours as variables in a Colors collection, typography as text styles, repeated shadows as shadow styles, and turns repeated cards and buttons into components. Variable names come from the site's own CSS custom properties where possible.
Can the imported page use my own design system's colours and styles?
Yes. If your file already has a variable or style with the same value, HTML to Frames reuses yours instead of creating a new one, and importing the same site again creates nothing new. With Snap to my file's styles on, colours and text are mapped to the closest styles in your own system even without an exact match, so a competitor or legacy page arrives already dressed in your system.
Do repeated cards and buttons become components in Figma?
Yes. In HTML to Frames Pro, frames that share the same structural fingerprint are grouped; the first becomes the main component and moves to a Components frame next to the page, and the others are replaced by instances. Differences in text, images and visibility are kept as overrides. If an instance doesn't match the original position and size, the original layer stays, so the design isn't broken.
Do button hover states come into Figma?
At widths of 1024 px and above, HTML to Frames Pro captures the hover state of links and buttons in bulk on the server using browser tooling. Elements whose look changes become a component set with Default/Hover variants, and a hover interaction is wired up in the prototype. Menus that open only via JavaScript on mouse hover aren't captured this way.
Do imported pages work as a clickable prototype?
Yes. HTML to Frames keeps links as Open link interactions, and when imported pages link to each other those links become Navigate to that frame. Fixed and sticky headers come with Figma scroll behaviour and horizontal carousels as scroll areas, so you can press Present and browse the site inside Figma.
What happens to collapsed content such as accordions or FAQs?
A collapsed details/summary block arrives in Figma in its collapsed state, showing only its heading. To keep the hidden answer text, HTML to Frames attaches it to the layer as a Figma annotation, or stores it as a hidden text layer where annotations aren't available.
Privacy, security and plans
Are the pages I import stored on the server?
No. In HTML to Frames page content isn't written to disk; the result is held in memory, delivered to the plugin once and deleted shortly after. Local file uploads are deleted as soon as the last page is processed and no public link is ever created. The plugin uses no cookies or tracking.
Does the Chrome extension send the content of my logged-in pages to a server?
In the copy-and-paste flow, page content never reaches a server; the capture goes to your clipboard or a .h2f file, and only the address and width are sent for the usage counter. With Send to Figma, the capture is held briefly in memory only to reach your open Figma plugin and is deleted the moment the plugin downloads it. Your password and cookies are never sent.
Does HTML to Frames read the contents of my Figma file?
No. HTML to Frames only writes to your file; the only thing it reads back is the source address and width it stored on frames it created itself, and it doesn't send that to the server. Your file's styles and variables are compared locally inside the plugin, purely for matching.
Is the capture service secure?
In HTML to Frames the browser that renders pages runs in an isolated container, separate from the API, with its sandbox enabled and no data of its own; image conversion runs in another container with no internet access. A firewall blocks private networks and server services, requests to internal addresses are refused, and abuse is met with rate limits and automatic blocking. Connections are encrypted over HTTPS.
Is HTML to Frames free? How do the plans work?
HTML to Frames has Free, Pro and Team plans. Free lets you try it with a monthly allowance of imports; Pro unlocks unlimited imports under fair use, up to four screens per import, custom widths, styles and variables, components and hover states, dark mode and more; Team offers the same features with a shared licence for your team. For current pricing and terms, contact Tati Creative via tatiajans.com.
How is an import counted? Are desktop, tablet and mobile separate?
In HTML to Frames one import is one page, including every screen you chose; Desktop + Tablet + Mobile together count as one import. A capture from the Chrome extension also counts as one import. The counter resets at the start of each month.
Can we use it with several people as an agency or team?
Yes, HTML to Frames' Team plan is designed for several people sharing one licence; the licence key is activated in each user's plugin. For a licence that fits your team size and for early access, write to Tati Creative via tatiajans.com.
Use cases
How do I bring an existing site into Figma for a redesign?
Starting a redesign by seeing the current state in Figma clarifies what needs to change and makes client conversations easier. With HTML to Frames you import the home page, pick the linked pages from the result screen and use the responsive set to get desktop, tablet and mobile side by side. That file, with Auto Layout and variables, becomes the direct starting point for the new design.
How do I bring competitor sites into Figma to analyse their design?
Competitor analysis needs pages side by side, measurable and annotatable. In HTML to Frames' Multiple mode you can import up to 10 competitor pages at once, compare each in its own row and read colour palettes, type scales and spacing systems from the generated variables and styles. The file is meant for internal analysis and inspiration, not for copying someone else's design one-to-one.
How do I prepare a site in Figma for a client presentation?
Agencies often want to show the current site and proposed changes side by side in a pitch. With HTML to Frames you import the client's site, design improvements right on top of it and, thanks to links between imported pages, present a clickable before/after prototype in Present mode.
As a developer, how do I hand a page I coded to a designer in Figma?
Screens built straight in code drift away from the design file over time. With HTML to Frames you can import a localhost page via the Chrome extension, or a built site as a .zip, and give the designer an up-to-date, editable Figma file, bringing design and code back in sync.
Can I import my WordPress, Shopify or Webflow site into Figma?
Yes. HTML to Frames doesn't care what platform a site is built on; it reads the final page as rendered in the browser. Any public page built with WordPress, Shopify, Webflow or custom code can be imported from its link, and CSS custom properties such as WordPress colour presets are reflected in variable names. For logged-in screens such as admin panels, use the Chrome extension.
Does it also convert Figma to a website (Figma to HTML)?
No, HTML to Frames works in one direction: it brings web pages and HTML into Figma. If you want a working website or online store built from a Figma design, that is part of Tati Creative's web design and development services; you can request a project via tatiajans.com.
御社での活用イメージを見てみませんか?
短いお打ち合わせでご要望を伺い、ライブデモをお見せします。導入と移行もお任せください。