# LandingRabbit homepage design playbook

## How to run this playbook

You are helping the user get a new homepage design for their business. When this file is uploaded, start Step 1 right away.

- Go through the steps in order. Finish each step before starting the next.
- Ask only the questions written in this playbook. Do not add, change, or invent questions.
- The user can answer by typing or by recording voice notes.
- Appendix B is used only in Step 4. Its steps belong to that task; the fonts, colours and design come from Steps 2 and 3.

---

## Step 1: Business answers

Ask the user to upload their latest sales materials, such as a sales deck, or to answer a few questions by talking to you.

- If the user uploads materials, check them against the questions below. If every question is answered in the materials, Step 1 is complete. Otherwise, ask only the questions the materials do not answer.
- If the user has nothing up to date, ask the questions below.

The questions:

1. If you are asked to describe your business in a few sentences, what do you say?
2. Who are your best customers?
3. What is the timing trigger for customers? How do they realize they need help?
4. What's your unique value proposition? Why should they choose you over another provider?
5. Do you have a step-by-step process for your service or is it fully custom per client?
6. How long is the typical engagement with a client?
7. If a customer is successful with you, how would they describe the transformation in their own words? What's better afterward?

---

## Step 2: Visual reference

Ask the user for an up-to-date visual reference: a slide deck, social media posts, brand guidelines, logo and fonts, or their existing website.

Then ask the user if they are happy with the main colours and fonts they'd like to use.

---

## Step 3: Homepage design

See the visual reference collected in Step 2.

Create a homepage design using the components from https://app.landingrabbit.com/design Use the visual reference but make sure the new look is modern and matches the user's answers from Step 2. Create an image that uses a variety of different sections from LandingRabbit design examples.

Choose a page structure that matches the customer the best, using the Step 1 answers and the page structure examples below.

Do not introduce anything in the image that is not included in the section's design example, and the section capabilities document in Appendix A includes detailed instructions per section. In all sections except navigation and footer, users can add a background image behind the text even if it's not shown in design examples. Navigation and footer aren't included in the design examples, but see https://landingrabbit.com for reference.

Every text element in every section has its own typography: font family, font size, font weight, line height, letter spacing, text transform, and colour.

Use the user's fonts. If you don't have access to a custom font, use the closest-matching Google Font.

The result is a design image of the full homepage.

If the user says anything looks wrong, review the design again against the section capabilities document in Appendix A and the design examples at https://app.landingrabbit.com/design, then fix it.

---

## Step 4 (optional): Brand guideline for LandingRabbit

When the design is done, ask the user:

"Would you like to create a live page with this design? I can create you a brand guideline file you can import into LandingRabbit in sign up and the app will create 70+ page components using this exact design."

If the user says yes:

Use the brand guidelines skill in Appendix B to create a design guideline for this homepage design. Make sure you cover every section and pay extra attention to defining each section's cta buttons and items separately. In every section and for every item, there are three cta buttons with rest and hover states. The guideline can only include flat and gradient section background colours, and background images need to be uploaded via the user interface after the workspace has been created. For the guideline, use a matching background colour. Include the mediaGrid section in the document.

Give the user the finished guideline as a Markdown file. Then tell the user to sign up at https://app.landingrabbit.com/signup and import the file there.

---

## Page structure examples

These well-known and tested landing page structure examples are for guidance. A homepage usually has every section of its brand-type homepage blueprint, in order, including its prose sections even when the materials contain no company story.

### Service homepage blueprint (agencies, clinics, consultancies, providers)

1. hero
2. trustedBy
3. features — label the offer, such as "Services" or "Programs"
4. mediaGrid — "Our work" where visitors judge the visible work or the place, layout media-grid; customized with layout alternate instead when supplied cases need substantial explanation
5. process — "How it works", only when the materials or the type of business has clear steps, method or process
6. testimonials
7. customized — label "About us", layout custom
8. faq
9. cta

### Organization homepage blueprint (nonprofits, care providers, coalitions, associations, public bodies)

1. hero
2. trustedBy
3. features — label "Our programs"
4. customized — label "Our mission", layout custom
5. customized — the ask to supporters, such as "Become a sponsor", layout custom
6. customized — label "About us" (who we are, history, partners), layout custom
7. cta

### Person homepage blueprint

1. hero
2. features — label "What I'm working on"
3. Examples of their work, such as "Projects":
   - When people judge the work by how it looks: a mediaGrid section, layout media-grid. Include it even when they haven't shared any work yet.
   - For other work: a customized section, only when the materials include examples. Layout alternate for two or three examples, boxes for more.
4. feed — label "Latest posts"
5. customized — label "About me" (who I am, background), layout custom
6. cta

### Product homepage blueprint

Typically PAS or AIDA.

PAS (Problem-Agitation-Solution):

1. hero
2. trustedBy
3. problem (challenge)
4. solution
5. features (services, products)
6. testimonials
7. faq
8. cta

AIDA (Attention-Interest-Desire-Action):

1. hero
2. trustedBy
3. benefits
4. features (services, products)
5. testimonials
6. faq
7. cta

---

# Appendix A: Section capabilities

# LandingRabbit section capabilities

One guide per section type, as the MCP tool manage_templates returns them in data.styling_capabilities.guide.

---

# Hero section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock hero section. The styling source of truth is
the hero editing instructions, layout tree, and CSS variable list that are
supplied separately to the planner.

## Layout structure

- There are three hero layouts:
  - horizontal: media beside text, media on the right
  - horizontal-reversed: media beside text, media on the left
  - vertical: text stacked above media, centered or left-aligned
- Every hero has one text column and one optional media slot (image, GIF,
  video, or embed). The slot can be hidden independently per desktop/mobile;
  fit is "fill" (crops to box) or "free-ratio" (natural ratio).
- The text column is an ordered stack of content elements. Elements can be
  added, removed, reordered, and repeated freely: eyebrow, h1, h2, h3, text
  (subtitle/paragraph), cta (a group of buttons), smallprint, boxes, and
  testimonial.
- boxes is a nested group of 2-4 sub-cards, each with icon, eyebrow, title,
  description, and CTAs. Column counts and card styling are CSS variables.
- testimonial is one quote with author name/role/organization, avatar, media,
  and a logo row.
- A video autoplays muted and loops by default (or shows a poster still, or
  opens in a lightbox); a GIF is an animated image; an embed is arbitrary code,
  so an animated or interactive visual can fill the media slot. None of these
  need custom design.
- In horizontal layouts the media is a bounded side column: it fills its grid
  column inside the content band, is sized by its own aspect ratio, and sits
  vertically centered. The text-to-media split, the column gap, and the content
  width are CSS variables.
- Buttons take a text label (free text; an emoji renders, no library icon) plus
  an optional trailing arrow, in primary/secondary/link variants. A button
  links to a URL or anchor, renders embed code inline or as a popup, or shows
  an inline search field.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- H-LIM-1: one media slot. Two side-by-side images or screenshots are not
  stock.
- H-LIM-2: no free or absolute positioning of foreground elements; the text
  column is a vertical stack. An overlapping foreground collage is not stock.
  Decorative shapes BEHIND content ARE stock — use the background layers.
- H-LIM-3: boxes sub-cards share one structure. Per-card bespoke markup is not
  stock.
- H-LIM-4: no parallax beyond background-attachment. A staggered entrance
  animation over the hero's elements is built in: the user toggles and tunes
  it in the section's layout settings in the editor, and it is set
  programmatically with the --lr-hero-animation-* variables; the element order
  it follows is derived from the layout and is not settable.
- H-LIM-5: no inline logo inside a text line (a "by [brand]" byline carrying
  the brand's logo). The eyebrow is text; the nearest stock is rendering the
  brand name as text.
- H-LIM-6: media never fills the section edge to edge or floor to ceiling. It
  sits inset in the padded content band — held off the top/bottom by the
  section padding and off the sides by the band's gutter — and that inset
  cannot be zeroed away. A full-bleed or full-height media half is not stock.
- H-LIM-7: hero boxes icons are stroke-based line glyphs chosen by name from
  the built-in library. Solid/filled, branded, uploaded, or exact badge icons
  are not reproducible; use the nearest line glyph from search_icons.
- H-LIM-8: the inner content card behind the text takes a solid color or
  gradient only, never an uploaded image, and the full-section background
  layers are rectangular. A rounded card showing a photo is not stock.
- H-LIM-9: a hero CTA button carries a text label plus an optional trailing
  arrow only. A library icon, inline glyph, or uploaded mark inside a button is
  not stock; an emoji in the label renders, otherwise the nearest stock is a
  text-only button.

## Styling

For styling effects, use the supplied hero editing instructions, hero layout
tree, and CSS variable list. Do not duplicate the variable reference here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include: five
stackable background layers (color, image, position, size, repeat, opacity,
z-index, each with mobile overrides), clip-path section edges, the inner
content card (background, padding, radius, shadow, border), per-element
typography, gradient headline text, lines above and below the headline, sub
headings and hero box titles, eyebrow pill styling, button variants with
hover states, adjacent-pair spacing and section padding, per-element max
widths, content width, the text-to-media split and column gap, image radius
and border, vertical-layout header alignment, and hero boxes card/icon
styling.

## MCP content contract

- Hero copy lives in content.elements, an ordered array. When elements is
  non-empty the renderer reads ONLY elements; top-level content.title,
  subtitle, ctas, and smallprint are legacy mirrors — never write them
  alongside elements.
- Element shapes: { type: "eyebrow", eyebrows: [{ text, variant }] };
  { type: "h1" | "h2" | "h3" | "text" | "smallprint", content: string };
  { type: "cta", ctas: [...] }; { type: "boxes", items: [...] };
  { type: "testimonial", testimonial: { quote, author: { name, position,
  organization } } }.
- Eyebrows are objects: { text: string, variant: "plain" | "pill" } inside the
  element's eyebrows array. A bare eyebrow string, or a singular eyebrow key,
  is not the MCP content shape.
- Hero boxes items use: id, eyebrow, title, description, ctas, and iconName.
  Use iconName from search_icons; never use icon. Do not use iconPosition;
  icon placement is controlled by the hero boxes icon layout CSS variables.
- Media (image/video/embed) is a user-supplied upload. The planner requests
  the asset instead of authoring it; hide an unused slot with the
  imageVisibility property and set fit with the imageFit property.
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "hero",
  sectionLayout "horizontal" | "horizontal-reversed" | "vertical", and
  section.content.elements using the shapes above.
- The hero layout of an existing section is changed with edit_page_styles
  change_layout, not through content.

---

# Trusted-by section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock trusted-by section. The styling source of
truth is the trusted-by editing instructions, layout tree, and CSS variable
list that are supplied separately to the planner.

## Layout structure

- Trusted-by has no layout variants: one centered vertical stack of content
  elements inside the section band.
- The content is an ordered stack of content elements. Elements can be added,
  removed, reordered, and repeated freely: eyebrow, h2 (the section title),
  h3, text (with inline links and bold), cta (a group of buttons), image,
  embed, and logos. Every part is optional. There is no h1 and no smallprint
  in a trusted-by.
- Only ONE logos element is allowed. It renders the logos as a single
  horizontal strip: a static centered row that wraps when space runs out
  (default), or a continuous auto-scrolling marquee. Scroll on/off, direction
  (right-to-left by default, or left-to-right), and speed (seconds per loop,
  5-120, default 90) are content fields on the logos element. The marquee
  repeats the logos to keep the loop seamless and falls back to the static
  wrapped row for reduced-motion visitors.
- Each logo is an uploaded image or video with an optional whole-logo link
  and a per-logo size scale, all managed in the editor. A logo has no
  caption, text label, or button.
- The image element takes an image, GIF, or video and the embed element takes
  arbitrary embed code; both are user-supplied, sit bounded inside the
  content band, and are sized by a 600/984/1440 preset width.
- Buttons take a text label (free text; an emoji renders, no library icon)
  plus an optional trailing arrow, in primary/secondary/link variants.
- A lightweight credibility band — a short label or heading over a row of
  brand logos — belongs here. A row of short TEXT items (feature ticks, a
  stat row, reassurance points) is not a logo strip: that is the boxes
  family. Quotes with author names are the testimonial section; a repeating
  list of full-width [media | text] rows is the alternate section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- TB-LIM-1: the logos render as ONE horizontal strip (a wrapped row or a
  marquee). A logo wall or grid with fixed rows and columns, logos at custom
  coordinates, or an overlapping collage is not stock; the nearest stock is
  the single strip, which wraps onto more lines only when space runs out.
- TB-LIM-2: logos are not CSS-styleable. Per-logo frames, plates, borders,
  shadows, or filters are not stock; a logo's appearance comes from the
  uploaded file, so a grayscale or faded strip needs grayscale/faded assets.
- TB-LIM-3: the gap between logos and the base logo row height are fixed.
  Only the per-logo size scale in the editor changes a logo's size.
- TB-LIM-4: the elements stack in one centered column. A side-by-side band
  ("Trusted by" text left, logos right on one line) or a multi-column
  arrangement of the elements is not stock; the nearest stock stacks the
  label above the strip.
- TB-LIM-5: logos cannot be interleaved with text, ratings, or badges in a
  composite row (logo + stat + logo). The nearest stock puts the text in its
  own element above or below the strip.
- TB-LIM-6: a logo takes an optional whole-logo link only. A caption, text
  label, or button under each logo is not stock, and rebuilding the strip as
  a boxes card grid is not an equivalent fallback: cards are not a logo
  strip and cannot auto-scroll (the marquee is trusted-by only).
- TB-LIM-7: the marquee is continuous and constant-speed, with only direction
  and seconds-per-loop adjustable. Hover-pause and a changed edge fade are not
  deliverable. Scroll entrance animation (section reveal, and a per-logo
  stagger that also runs in marquee mode) is built in: the user toggles and
  tunes it in the section's layout settings in the editor, and it is set
  programmatically with the --lr-trusted-by-animation-* variables.

## Styling

For styling effects, use the supplied trusted-by editing instructions,
trusted-by layout tree, and CSS variable list. Do not duplicate the variable
reference here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include: the
section background (color, gradient, or image band), clip-path section edges,
content width, top and bottom section padding, eyebrow pill styling,
per-element typography for the h2, h3, and text elements, lines above and
below the h2 and h3, text link and bold
styling, and button variants with hover states.

## MCP content contract

- Trusted-by copy lives in content.elements, an ordered array. There are no
  top-level content fields. Element shapes:
  { type: "eyebrow", eyebrows: [{ text, variant }] };
  { type: "h2" | "h3" | "text", content: string };
  { type: "cta", ctas: [...] };
  { type: "logos", logosScroll, logosScrollDirection, logosScrollSpeed }.
- Eyebrows are objects: { text: string, variant: "plain" | "pill" } inside the
  element's eyebrows array. A bare eyebrow string, or a singular eyebrow key,
  is not the MCP content shape.
- The scroll settings sit on the logos element: logosScroll (boolean),
  logosScrollDirection ("right-to-left" | "left-to-right"), and
  logosScrollSpeed (seconds per loop, 5-120). They are content fields, not
  update_section_property toggles.
- Logo entries are user uploads managed in the editor (add or swap, per-logo
  scale, per-logo link). Never author or delete logo entries: add_section
  seeds the section from the workspace's most recent trusted-by section
  (logos and scroll settings carry over), and on update_section the logos
  element is reused by id with no logos key so the existing entries survive.
- Image and embed elements are user-supplied. The planner requests the asset
  or embed code instead of authoring it.
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "trustedBy" and
  section.content.elements using the shapes above.
- Trusted-by has a single layout; there is no sectionLayout and no
  change_layout step.

---

# Boxes section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock collection boxes section. The styling source
of truth is the boxes editing instructions, layout tree, and CSS variable list
that are supplied separately to the planner.

## Layout structure

- There are three collection boxes layouts:
  - boxes: grid items with one image/GIF/video media slot
  - boxes-icons: grid items with one built-in line icon
  - boxes-text: text-only grid items
- Every boxes section has one optional header and one grid of items.
- The section header is a stack of any/all of: eyebrow, title, subtitle, and CTA
  buttons.
- By default, the header sits above the grid and is centered. The header stack
  can be left-aligned while staying above the grid. It can also move to a left
  side column, with the grid on the right.
- Grid items can include any/all of the supported content slots listed in the
  MCP content contract. Items can be reordered by changing content order.
- Per item, the content is one ordered stack of optional parts: media or icon
  (by layout), eyebrow, title, text, embed, nested cards (smaller cards, each
  with eyebrow, icon, title, text and buttons), and CTA buttons. The item as a
  whole can optionally link to one URL. When it does, the item shows as
  clickable on the published page, and buttons and links inside it keep their
  own targets.
- In boxes, the media slot can appear at the top or bottom of each item. A card
  cannot have both an image/media slot and an icon slot.
- In boxes-icons, the icon can sit above the title/text or to the left of the
  title/text. This is a styling change, not content data: use the icon layout
  CSS variables from the styling instructions. Do not use iconPosition for
  collection boxes.
- Grid column count and grid alignment can be altered with styling. Items remain
  one shared size and one shared layout kind per grid.
- A grid cannot mix image cards, icon cards, and text-only cards. It also cannot
  mix item sizes, spans, masonry, or bento-style cards. Those require custom
  design.
- Grid items can have dividers between them, or the section can use a manual
  carousel with left/right controls. The carousel is not auto-scrolling.
- Alternate full-width rows with media beside text are a different collection
  layout and are not covered by this boxes planner.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- G-LIM-1: the section has one header and one content region. Side labels,
  sub-group labels, split headers, and separate panels between header and grid
  are not part of the boxes layout.
- G-LIM-2: the header is a stack. It can be above the grid or in a left side
  column, but arbitrary header placement is not supported.
- G-LIM-3: the section has one background across the whole band. Cards have
  their own backgrounds/borders, but the grid region does not have a separate
  background panel, and there is no stock separator rule between header and grid.
- G-LIM-4: geometry follows the stock section flow. Spacing, widths, columns,
  and alignment can change through styling, but extra layout primitives such as
  absolute-positioned decoration, overlap, or independent panels are not stock.
- C-LIM-1: all items in a grid share one structure. Per-item overrides can
  restyle existing slots, but not create bespoke markup per card.
- C-LIM-2: items sit in a uniform grid. No free positioning, overlap, rotation,
  z-index layering, masonry, bento, mixed widths, or column spans.
- C-LIM-3: each card has at most one media/icon slot. A card needing image plus
  icon, multiple images, or a custom logo inside the icon slot is not stock.
- C-LIM-4: icon glyphs come from the built-in stroke icon library. Solid/filled,
  duotone, branded, uploaded, or exact certification badge icons are not
  reproducible in the icon slot; use the nearest line glyph from search_icons.
- C-LIM-5: the icon badge is shaped with radius only. Circle, square, and
  rounded square are stock; rotated diamond/rhombus frames are not.
- C-LIM-6: media is a discrete slot above or below text. Full-bleed photo cards
  with text overlaid are not stock.
- C-LIM-7: carousel is manual only. Auto-scroll, timed sliding, and parallax
  are not deliverable with stock boxes. Scroll entrance animation (section
  reveal and item appear) is built in: the user toggles and tunes it in the
  Animation tab of the section's layout settings in the editor, and it is set
  programmatically with the --lr-boxes-animation-* variables.
- C-LIM-8: hover applies to items that have a whole-item link, or to every
  item when hover-items is all. It is one motion (none, lift, grow), and the
  hovered item can take its own background colour or gradient, border colour
  and shadow, which combine with the motion. Every part of the item can be
  switched on to change with it:
  eyebrow, title, text, image or icon (by layout), and the button (its arrow
  nudges with it). Each part has a built-in look that its own hover variables
  can tune. It is built in: the user tunes it under Items on hover in the
  Animation tab, and it is set programmatically with the
  --lr-boxes-animation-*-hover-* variables. Other hover behaviour (per-item
  effects, revealing hidden content) is not stock.

## Styling

For styling effects, use the supplied boxes editing instructions, boxes layout
tree, and CSS variable list. Do not duplicate the variable reference here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include:
section/header-to-grid geometry, section padding, content width, grid columns,
grid alignment, card gap, icon above-vs-left layout, icon gap/alignment, icon
size/stroke/badge styling, card backgrounds, borders, shadows, radius, text
colors, typography, CTA styling, per-item styling, lines above and below the
section title and card titles, dividers, and carousel.

## MCP content contract

- For add_section/update_section, collection boxes content must use renderable
  keys only. Supported section content keys are: eyebrow, title, subtitle, ctas,
  bottomCtas, and items. Do not emit overline, kicker, tagline, heading, or icon.
- Section and item eyebrows are objects: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- Collection boxes items use: id, eyebrow, title, description, image, media,
  imageLink, link, openInModal, disableAutoloop, hideControls, posterImage, embed,
  ctas, contentItems, boxes, testimonial, and iconName. For boxes-icons, use
  iconName from search_icons; never use icon. Do not use iconPosition for
  collection boxes; moving icons left/above is controlled by CSS variables.
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "collection",
  sectionLayout "boxes" | "boxes-text" | "boxes-icons", and section.content
  using the keys above.

---

# Alternate section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock collection alternate section. The styling
source of truth is the alternate editing instructions, layout tree, and CSS
variable list that are supplied separately to the planner.

## Layout structure

- The collection alternate layout is a repeating list of full-width rows. Each
  row is one item: a two-column [media | text] split, with the media side
  alternating per row (first row media left by default).
- Every alternate section has one optional header and one list of item rows.
- The section header is a stack of any/all of: eyebrow, title, subtitle, and
  CTA buttons. It sits above the items, centered by default, and can be
  left-aligned while staying above the items. Unlike boxes, it cannot move to
  a side column.
- Each row spans the full content width. Rows cannot sit two or more side by
  side and cannot vary in width. The media-to-text split ratio, the column
  gap, and the alternating order (default alternating, all left, all right,
  or first row right) are styling changes, including the mobile order.
- Per item, the text column is a stack of: eyebrow, title, description, and
  CTA buttons. The whole row can optionally link to one URL; buttons and links
  inside it keep their own targets. An item can additionally nest a boxes sub-card grid, one
  testimonial, and embed content items. Items can be reordered by changing
  content order.
- The media slot per item takes an image, GIF, video, or embed. It is its own
  column BESIDE the text — never inside a card with the text and never behind
  it. Media fit is a data toggle: "fill" (crops to the column) or
  "free-ratio" (natural ratio).
- There is no item-level icon slot. Icons appear only inside nested boxes
  sub-cards, chosen by name from the built-in line icon library.
- Item rows can be plain (no container) or carry a shared box treatment
  (background, radius, border, shadow) through styling, with per-item
  overrides by position.
- Alternate is a repeating list: a single media-beside-text block has nothing
  to alternate, so one full-height image beside text belongs to a hero, steps,
  or cta section, not a one-item alternate.
- A uniform grid of tiles/cards with the media INSIDE each card above or
  below its caption is the boxes family and is not covered by this alternate
  planner.
- One media column beside a vertical list of short icon-led rows is the steps
  section, not alternate. Steps repeats ONE shared decoration glyph for every
  row; it has no per-row icons.
- There is no dividers toggle and no carousel for alternate; those belong to
  the boxes layouts.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- G-LIM-1: the section has one header and one content region. Side labels,
  sub-group labels, split headers, and separate panels between header and rows
  are not part of the alternate layout.
- G-LIM-2: the header is a stack above the items. A side-column header or
  arbitrary header placement is not supported.
- G-LIM-3: the section has one background across the whole band. Item rows
  can carry their own box background/border/shadow, but the rows region has
  no separate background panel, and there is no stock separator rule between
  header and rows or between rows.
- G-LIM-4: geometry follows the stock section flow. Spacing, widths, the
  column split, and alignment can change through styling, but extra layout
  primitives such as absolute-positioned decoration, overlap, or independent
  panels are not stock.
- A-LIM-1: all items share one structure and one style-variable set. Per-item
  overrides can restyle existing slots, but not create bespoke markup per row.
- A-LIM-2: one item per full-width row. Two or more items side by side, mixed
  row widths, or a grid of horizontal media-beside-text cards is not stock.
- A-LIM-3: each item has at most one media slot and no item-level icon slot.
  An item needing two images, or an icon/number leading each row's text, is
  not stock; the nearest stock for an icon-led list beside one media column is
  the steps section, which repeats one shared glyph for every row rather than
  a distinct icon per row.
- A-LIM-4: media is a bounded column inside the padded content band, beside
  the text. Full-bleed media, media behind the text, or text overlaid on the
  media is not stock.
- A-LIM-5: nested boxes sub-cards share one structure, and their icon glyphs
  come from the built-in stroke icon library. Solid/filled, duotone, branded,
  uploaded, or exact badge icons are not reproducible; use the nearest line
  glyph from search_icons.
- A-LIM-6: no parallax beyond background-attachment, and no auto-scroll or
  timed sliding. Scroll entrance animation (section reveal and item appear)
  is built in: the user toggles and tunes it in the Animation tab of the
  section's layout settings in the editor, and it is set programmatically with
  the --lr-alternate-animation-* variables.
- A-LIM-7: hover applies to rows that have a whole-row link, or to every row
  when hover-items is all. It is one motion (none, lift, grow), and the hovered
  row can take its own background colour or gradient, border colour and
  shadow, which combine with the motion. Every part of the row can be switched
  on to change with it: eyebrow,
  title, text, media, nested boxes, testimonial, and the button (its arrow
  nudges with it). Each part has a built-in look that its own hover variables
  can tune. It is built in: the user tunes it under Items on hover in the
  Animation tab, and it is set programmatically with the
  --lr-alternate-animation-hover-* variables. Other hover behaviour (per-row
  effects) is not stock.

## Styling

For styling effects, use the supplied alternate editing instructions,
alternate layout tree, and CSS variable list. Do not duplicate the variable
reference here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include:
section padding, content width, header alignment and the header-to-items gap,
the gap between rows, the media-to-text column split, the alternating order
(odd/even and mobile), item box backgrounds/borders/shadows/radius/padding,
image radius and border, text colors, typography, eyebrow pills, CTA styling
(section, item, and bottom), nested boxes card/icon styling, testimonial
styling, lines above and below the section title, item titles and nested
box titles, and per-item styling.

## MCP content contract

- For add_section/update_section, collection alternate content must use
  renderable keys only. Supported section content keys are: eyebrow, title,
  subtitle, ctas, bottomCtas, and items. Do not emit overline, kicker,
  tagline, heading, or icon.
- Section and item eyebrows are objects: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- Collection alternate items use: id, eyebrow, title, description, image,
  media, imageLink, link, openInModal, disableAutoloop, hideControls, posterImage,
  embed, ctas, contentItems, boxes, and testimonial. There is no item-level
  iconName. Nested boxes sub-card items use: id, eyebrow, title, description,
  and iconName from search_icons; never use icon or iconPosition.
- Media (image/video/embed) is a user-supplied upload. The planner requests
  the asset instead of authoring it; media fit is set with the imageFit
  property ("fill" | "free-ratio").
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "collection",
  sectionLayout "alternate", and section.content using the keys above.
- The layout of an existing collection section is changed with
  edit_page_styles change_layout, not through content.

---

# Testimonial section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock testimonial section. The styling source of
truth is the testimonial editing instructions, layout tree, and CSS variable
list that are supplied separately to the planner.

## Layout structure

- There are two testimonial layouts:
  - carousel: one card visible at a time, with prev/next arrows and optional
    timed autoplay
  - grid: a multi-column wall of cards
- The layout is section data, not styling. Every card shares one structure in
  both layouts; a section cannot mix layouts or vary the arrangement card to
  card.
- Every testimonial section has one optional header, the cards, and optional
  bottom CTA buttons after the cards.
- The section header is a stack of any/all of: eyebrow, title, subtitle, and
  CTA buttons. By default it sits above the cards and is centered; it can be
  left-aligned while staying above the cards, or move into a LEFT or RIGHT
  side column beside the cards via a data toggle. The side column split is a
  styling change.
- Each card is one fixed ordered stack, every part optional except the quote:
  a client logo row (one or more uploaded logos), the quote, and an author
  block (avatar, name, position, organization). The quote is rich text:
  inline links and bold are stock, with their own styling variables per text
  role (quote, author, position, organization).
- A card can additionally carry one media slot (image, GIF, video, or embed).
  In the carousel the media renders as a large panel on the LEFT of the card,
  about 40% of the card width; in the grid it sits on TOP of the card above
  the text. When the header is in a side column, carousel media moves on top
  of the card instead of beside it.
- In the carousel only one card is visible at a time; the arrows appear when
  there is more than one card. Autoplay is a data toggle with a duration of
  2-30 seconds (default 5); it cross-fades between cards and pauses on
  hover.
- In the grid the cards are equal height, with the author block pinned to the
  card bottom. The desktop column count adapts to the card count (two or
  three columns) and can be overridden with styling; with a side header the
  grid is capped at two columns.
- Carousel text is centered by default; grid text is left-aligned by default.
  Both are styling changes, as are avatar and logo alignment.
- Buttons take a text label (free text; an emoji renders, no library icon)
  plus an optional trailing arrow, in primary/secondary/link variants.
- A quote wall or a one-at-a-time quote rotator belongs here. One quote
  inside a hero or CTA stack is that section's testimonial element; a logo
  strip with no quotes is the trustedBy section; a uniform grid of feature
  cards is the boxes family.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- TST-LIM-1: all cards share one stock structure. A featured or unique card,
  or cards that differ from one another in arrangement, is not stock; the
  nearest stock gives every card the same structure.
- TST-LIM-2: the cards sit in one uniform grid or one single-card carousel.
  Masonry or staggered-height packing, mixed column widths, and a multi-card
  sliding carousel (several cards moving together, or adjacent cards peeking
  in at the edges) are not stock.
- TST-LIM-3: the card order is fixed: logos, then quote, then author block
  with the avatar beside the name/position/organization stack. Reordering
  these parts or free-positioning them inside the card is not stock; only
  alignment and gaps are styling changes.
- TST-LIM-4: card framing (background, border, radius, shadow) is one
  section-wide value. A single highlighted card with its own background,
  border, or accent color is not stock.
- TST-LIM-5: a testimonial item carries only logos, quote, media, and author
  name/position/organization. Structured per-card fields — star ratings,
  score badges, source-platform chips — have no slot; the nearest stock
  omits them.
- TST-LIM-6: the media panel position is fixed: left in the carousel (about
  40% of the card width), on top in the grid. A right-side panel, a changed
  split, full-bleed media, or text overlaid on the media is not stock.
- TST-LIM-7: no parallax, and in-place motion is limited to the carousel
  arrows and its cross-fade autoplay. Scroll entrance animation (section
  reveal and item appear) is built in: the user toggles and tunes it in the
  section's layout settings in the editor, and it is set programmatically with
  the --lr-testimonial-animation-* variables; in the carousel, item appear
  reveals the carousel as one unit because only one testimonial is visible at
  first render.
- TST-LIM-8: logos and avatars are uploaded images and there is no icon slot
  anywhere in the section. A library glyph, decorative quote mark, or
  generated logo cannot stand in for an upload; the nearest stock leaves the
  slot empty until the user supplies the asset.

## Styling

For styling effects, use the supplied testimonial editing instructions,
testimonial layout tree, and CSS variable list. Do not duplicate the variable
reference here.

Many variables are layout-scoped: grid-prefixed variables apply only to the
grid layout and carousel-prefixed variables only to the carousel, while card
radius, shadow, and border are shared. A styling fact is achievable only when
it can be mapped to an existing slot, toggle, or CSS variable from those
inputs. Examples to observe include: section background and clip-path edges,
section padding, content width, header alignment and max width, lines above
and below the section title, the header-to-cards gap, the side-header column split, card background, radius,
shadow, border, and padding (with grid mirrors), grid columns and item max
width, the carousel element and quote widths, quote and author
name/position/organization typography with per-role inline link and bold
styling, avatar size/shape/gap and alignment, logo height/gap/margin and
alignment, quote-to-author spacing, text alignment per layout, carousel
arrow colors, eyebrow pill styling, button variants with hover states for
header and bottom CTAs, and bottom button alignment.

## MCP content contract

- For add_section/update_section, testimonial content must use renderable
  keys only. Supported section content keys are: eyebrow, title, subtitle,
  ctas (the header buttons), items, and bottomCtas (the buttons after the
  cards). Do not emit overline, kicker, tagline, heading, icon, or elements.
- The section eyebrow is an object: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- Testimonial items use: id, quote, logos, media, image, imageLink,
  openInModal, disableAutoloop, hideControls, posterImage, embed, and
  author. The quote may carry simple inline HTML links and bold. Every item
  requires author: { name, position, organization } — use empty strings for
  unknown fields; author.image is the avatar. There is no item-level title,
  description, iconName, or ctas key.
- Media (card image/video/embed), logos, and avatars are user-supplied
  uploads. The planner requests the assets instead of authoring them; media
  fit is set with the imageFit property ("fill" | "free-ratio").
- The header placement is section data, not content: sectionHeaderPosition
  is "above", "left", or "right". On add_section it sits on the section
  object beside name and content; on an existing section it is changed with
  edit_page_content update_section_property, never through content.
- Autoplay is section data, not content, and applies to the carousel layout
  only: carouselAutoplay is a boolean and carouselAutoplayDuration is
  seconds from 2 to 30 (default 5). On add_section they sit on the section
  object beside name and content; on an existing section they are changed
  with edit_page_content update_section_property, never through content.
- When the planner outputs contentPlan for add_section, it should be
  directly translatable to edit_page_content arguments: sectionType
  "testimonial", sectionLayout "carousel" | "grid", and section.content
  using the keys above.
- The testimonial layout of an existing section is changed with
  edit_page_styles change_layout, not through content.

---

# Media grid section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock media grid section. The styling source of
truth is the media grid editing instructions, layout tree, and CSS variable
list that are supplied separately to the planner.

## Layout structure

- The section is one ordered stack. Every part is optional and renders only
  under the condition stated:
  1. Section header — renders when any of eyebrow, title, subtitle, or header
     CTA buttons carries content. It is a stack of those four, in that order.
     It sits above the tiles, centered or left-aligned by styling. A data
     toggle moves it to a side column on the left of the tiles.
  2. Tiles — renders when there is at least one item. Every tile is mainly its
     media filling the tile: one image, GIF, or video, or an embed.
  3. Bottom CTA — renders when bottom CTA buttons carry content. Buttons sit
     below the tiles.
- Four layouts arrange the tiles:
  - media-grid: even rows of same-shaped tiles. The column count and the tile
    shape (an aspect ratio, a minimum that grows with text) are styling.
  - media-mosaic: a bento of tiles sized by position as small, wide, tall, or
    large. Without per-tile sizes a built-in rhythm fills the grid. Column
    count and row height are styling.
  - media-carousel: one row of tiles that scrolls sideways with arrow buttons.
    Card width is styling. Optional auto-scroll moves slide by slide every N
    seconds (2 to 30) or as a continuous loop in either direction at a set
    speed. The continuous loop runs only when the tiles overflow the row by
    more than one tile, and stops for good once a visitor presses an arrow or
    tabs into it. Auto-scroll pauses on hover and focus.
  - media-expand: a row of tall panels; the open panel widens. With text on
    hover, the open panel's text appears on hover or focus and closed panels
    show nothing until hovered or focused; with text always visible, the open
    panel shows its text and closed panels show only a vertical title, or
    nothing when the tile has none or its title does not fit the panel. The first panel starts open; hover or a click opens another on
    desktop, a tap on phones, and a key press anywhere. Panels stack as an
    accordion on phones. Optional autoplay opens the next panel every N seconds
    (2 to 30), on desktop only, and stops for good once a visitor opens a
    panel.
- Neither carousel auto-scroll nor expand autoplay runs for visitors who
  reduce motion.
- One tile is its media with an ordered stack of optional parts on top of it:
  eyebrows, a sub heading, text, and CTA buttons, over a shade that keeps the
  text readable. The shade can take its own colour on hover. Which parts render
  is decided by which items carry content.
  The tile as a whole can link to one URL; buttons inside keep their own
  targets.
- Every tile reacts to the item hover effect; there is no linked-only option.
- In all four layouts a data toggle decides whether tile text and buttons show
  only on hover or all the time. Touch screens and keyboard focus always show
  them. In media-expand, on hover closed panels show nothing until hovered or
  focused; always visible shows the open panel's text and only the title on
  closed panels. With text always visible, the text blocks of tiles in one row
  start at the same height and their buttons line up at the bottom.
- Tile media always fills its tile. In media-expand it keeps the open panel's
  size, so opening a panel shows more of it without zooming, for images and
  videos alike.
- In media-mosaic a tile's column span and row span are each its own saved
  value, else the value saved on the matching tile among the first six (tile 7
  and tile 13 match tile 1), else a section-wide value, else the built-in
  large, tall, small, small, wide, wide pattern. Tiles then adjust so the grid
  has no gaps and no tile is wider than the columns, for any column count on
  desktop and mobile. Saved spans are 1 or 2; the grid may grow the last tiles
  past that to close gaps.
- Text-led cards with a smaller image beside or above the text are the boxes
  family; content the reader clicks between is the tabs section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- MEDIA-GRID-LIM-1: tile text always sits on the media, over the shade. Text
  below or beside the media, outside the tile, is the boxes family.
- MEDIA-GRID-LIM-2: the shade cannot go lighter than the strength that keeps
  the tile text readable over any photo. Unshaded photos with text are not
  stock; showing text only on hover is the nearest fallback.
- MEDIA-GRID-LIM-3: one tile holds one media slot: an image, GIF, video, or
  embed. A slideshow of several images in one tile, two media side by side
  inside one tile, an image plus an icon, or a tile without media that is
  meant to stay empty is not stock.
- MEDIA-GRID-LIM-4: a mosaic tile's saved size spans one or two columns and one
  or two rows. The grid may widen or lengthen the last tiles to close gaps.
  Choosing larger sizes, free positioning, and overlapping tiles are not stock.
- MEDIA-GRID-LIM-5: a section uses one layout for all its tiles. Mixing even
  grid rows with a mosaic, or a carousel with expanding panels, is not stock.
- MEDIA-GRID-LIM-6: the carousel scrolls one row sideways. Vertical scrolling,
  several rows, dots or thumbnails as navigation, and a per-tile duration are
  not stock.
- MEDIA-GRID-LIM-7: expanding panels sit in one horizontal row on desktop. A
  vertical stack on desktop, more than one open panel, and panels that open on
  click only are not stock.
- MEDIA-GRID-LIM-8: a full-size view on click opens the tile's image in the
  standard modal. A lightbox gallery with its own navigation is not stock.

## Styling

For styling effects, use the supplied media grid editing instructions, layout
tree, and CSS variable list. Do not duplicate the variable reference here.

A styling fact is achievable only when it can be mapped to an existing
slot, toggle, or CSS variable from those inputs. Examples to observe
include: section background and padding, content width, the header's
alignment and widths, the gap between the header and the tiles and between
tiles, grid columns and tile shape, mosaic columns, row height and per-tile
spans, carousel card width and arrow buttons, expand panel heights and the
closed-panel title, tile radius, border and shadow, the shade colour, its
hover colour and its fade, the tile text block's padding and alignment, tile eyebrow, heading,
text, and button styling, lines above and below the section title and tile
headings, hover effects, and bottom button variants.

## MCP content contract

- For add_section/update_section, media grid content must use renderable keys
  only. Supported section content keys are: eyebrow, title, subtitle, ctas (the
  header buttons), bottomCtas, and items. Do not emit overline, kicker,
  tagline, heading, or elements.
- Section eyebrows are objects: { text: string, variant: "plain" | "pill" }. A
  bare eyebrow string is not the MCP content shape.
- A tile item is { id, contentItems } plus its media fields (image, media,
  link, openInModal, disableAutoloop, hideControls, posterImage, embed). link
  covers the whole tile; a tile has no separate image link.
  contentItems is an ordered array; each entry has id and type, where
  type is one of eyebrow, h3, text, or cta, plus the field that type carries:
  eyebrows (an array of the eyebrow object) for eyebrow, content for h3 and
  text, ctas for cta.
- The layout is section data, not content: layout is "media-grid",
  "media-mosaic", "media-carousel", or "media-expand" and sits on the section
  object beside name and content on add_section; on an existing section it is
  changed with edit_page_styles change_layout.
- The tile and motion settings are section data too, changed with
  edit_page_content update_section_property: mediaGridTextVisibility ("hover"
  or "always"; all four layouts),
  mediaGridCarouselAutoScroll ("off", "slide", or "continuous"),
  mediaGridCarouselSeconds (2 to 30), mediaGridCarouselDirection
  ("right-to-left" or "left-to-right"), and mediaGridCarouselSpeed (seconds per
  loop) on media-carousel; mediaGridExpandAutoplay (boolean) and
  mediaGridExpandSeconds (2 to 30) on media-expand; sectionHeaderPosition
  ("above" or "left"; all four layouts).
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "mediaGrid",
  sectionLayout one of the four layouts, and section.content using the keys
  above.

---

# Pricing section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock pricing section. The styling source of
truth is the pricing editing instructions, layout tree, and CSS variable list
that are supplied separately to the planner.

## Layout structure

- Pricing has one layout: a uniform responsive grid of plan cards. There are
  no layout variants.
- The section is an ordered stack of three regions: one optional header, one
  optional period/currency selector row, and one card grid. The region order
  is fixed.
- The section header is a stack of any/all of: eyebrow, title, subtitle, and
  CTA buttons. It sits above the cards, centered by default, and can be
  left-aligned while staying above the cards; that alignment is a styling
  change. It cannot move to a side column.
- The section ends at the card grid. There is no bottom region: no bottom
  text and no bottomCtas.
- Each card is an ordered stack of content rows. Rows can be added, removed,
  reordered, and repeated freely: eyebrow (the plan name or a "Most popular"
  badge), price, h3 (a heading), subtitle (a subheading), paragraph (rich
  text; the feature list), and cta (a group of buttons). Every row is
  optional and renders only when its content is non-empty.
- The price row is one baseline: the amount with its currency before or
  after it, then the period and the unit, each after a "/" divider. The
  price row and the period/unit have separate typography; the amount alone
  cannot be styled independently of the row. An amount without digits
  ("Custom") renders as-is.
- Feature lists live inside paragraph rows as bulleted lists. Each list
  carries one built-in glyph type — default dot, check, star, arrow,
  success, or destructive — recolorable by type; different lists in one card
  can use different types. Library icons, uploaded marks, and per-row custom
  glyphs are not available.
- Content rows align across the grid automatically when every card shares
  the same ordered row-type structure, so prices and buttons line up; a card
  whose structure differs flows freely instead. To line rows up, plan
  matching row order and count on every card.
- The grid picks its desktop column count from the card count (1 card → 1
  column, 2 → 2, 4/8/12 → 4, any other count → 3 with extra cards wrapping
  to new rows; two columns on tablet, one on mobile).
  Column count and the shared card max width are styling changes. Cards
  share one size and one style-variable set; per-card overrides by position
  restyle existing rows.
- The period/currency selector renders only when the section has billing
  period or currency groups configured. It shows as tab groups by default; a
  two-option group can show as an on/off switch — the switch style is chosen
  by the user in the editor. Groups, their options, and per-period prices
  are configured in the editor, not through this surface.
- Featured-plan emphasis is per-card styling by position (background,
  border, shadow, radius, padding) plus an eyebrow badge row. There is no
  ribbon or badge element and no card scaling.
- There is no media or icon slot anywhere in the section. Buttons take a
  text label (free text; an emoji renders, no library icon) plus an optional
  trailing arrow, in primary/secondary/link variants.
- A uniform grid of non-pricing cards is the boxes family; a two-column
  before/after contrast is the comparison section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- G-LIM-1: the section is one header, one optional selector row, and one
  card grid, in that order. Side labels, split headers, and separate panels
  between the regions are not part of the pricing layout.
- G-LIM-2: the header is a stack above the cards. A side-column header or
  arbitrary header placement is not supported.
- G-LIM-3: the section has one background across the whole band. Cards carry
  their own backgrounds/borders, but the grid region has no separate
  background panel, and there is no stock separator rule between the
  regions.
- G-LIM-4: geometry follows the stock section flow. Spacing, widths,
  columns, and alignment can change through styling, but extra layout
  primitives such as absolute-positioned decoration, overlap, or independent
  panels are not stock.
- PRI-LIM-1: a feature-comparison matrix — feature rows × plan columns with
  per-cell marks, row separators spanning the grid, or a feature-label
  column — is not stock. The nearest stock is a bulleted feature list per
  card; for exactly two plans the comparison section MAY fit.
- PRI-LIM-2: all cards share one content model: ordered rows of the six row
  types. Per-card overrides by position restyle existing rows, but bespoke
  markup inside a card is not stock.
- PRI-LIM-3: featured-plan emphasis is per-card background, border, shadow,
  radius, and padding plus an eyebrow badge row. A ribbon, corner fold,
  diagonal banner, a scaled-up or physically taller card, or a card
  overlapping its neighbours is not stock.
- PRI-LIM-4: the selector is the built-in tab groups or the two-option
  on/off switch, and it sits between the header and the cards. The switch
  style is chosen in the editor, not through this surface. A segmented
  slider, dropdown, icon pills, or a selector anywhere else is not stock.
- PRI-LIM-5: feature rows are rich text with one glyph type per list,
  recolorable by type. Per-feature tooltips, info icons, expandable rows,
  library icons per row, or annotated cells are not stock.
- PRI-LIM-6: the price row is amount, currency, period, and unit. A
  struck-through compare-at price, a savings badge inside the price row, or
  an amount styled independently of the row is not stock; the nearest stock
  is a separate eyebrow or paragraph row.
- PRI-LIM-7: the period/currency switcher and per-period prices come from
  groups configured in the editor. When a design shows a Monthly/Yearly or
  currency switcher, plan the visible default prices and name the switcher
  as an editor follow-up; the selector's own styling is stock once the
  switcher exists.
- PRI-LIM-8: no media or icon slot: plan logos, illustrations, photos, or
  icon-led columns inside cards are not stock. An emoji in text renders; the
  nearest stock is text-only rows.
- PRI-LIM-9: the section ends at the card grid. A closing line of text or
  buttons under the grid is not a pricing element; the nearest stock is the
  header CTAs or a separate cta section.
- PRI-LIM-10: no parallax beyond background-attachment. Scroll entrance
  animation (section reveal and item appear) is built in: the user toggles
  and tunes it in the section's layout settings in the editor, and it is set
  programmatically with the --lr-pricing-animation-* variables.

## Styling

For styling effects, use the supplied pricing editing instructions, pricing
layout tree, and CSS variable list. Do not duplicate the variable reference
here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include:
section background and padding, content width, header alignment and max
width, lines above and below the section title, the gap between the section header and the pricing body, selector bar
alignment, order, and spacing, tab container and tab button styling, switch
track/thumb/label colors, grid column overrides, the gap between cards (row
and column), card max width, card backgrounds, borders, shadows, radius,
content padding, per-row typography (plan-name eyebrow, price, period/unit,
headings, subheadings, feature text), feature-glyph colors by type (card
lists and the header subtitle list are separate scopes), eyebrow pills,
section and per-card button variants with hover states, and per-card styling
by position.

## MCP content contract

- For add_section/update_section, pricing content must use renderable keys
  only. Supported section content keys are: eyebrow, title, subtitle, ctas
  (the header buttons), and items (the plan cards). There is no bottomCtas.
  Do not emit overline, kicker, tagline, heading, icon, or elements. Never
  emit groups, customCurrencies, or item contentVariants — the
  period/currency switcher and its per-period content are editor-managed.
- Section and row eyebrows are objects: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- items holds the plan cards in order: { id, contentItems }. contentItems is
  the ordered row list per card. Row shapes:
  { type: "eyebrow", eyebrow: { text, variant } };
  { type: "price", price: { amount, currency, currencyPosition: "before" |
  "after", period, unit } };
  { type: "h3" | "subtitle" | "paragraph", content: string };
  { type: "cta", ctas: [...] }. Each row also carries its id.
- Price parts are plain text without separators: the renderer places the
  currency and adds a "/" before the period and before the unit. Use empty
  strings for parts the design does not show. currencyPosition defaults to
  "after" for € and "before" otherwise; set it explicitly when the design
  shows it.
- paragraph content is rich text: inline links and bold are stock. A feature
  list is written as plain text lines inside one paragraph row: a leading
  ✅/✓ (success), ☑ (check), ❌/✗ (destructive), ⭐ (star), or → (arrow)
  marker chooses that line's glyph and is removed from the text; "- " lines
  take the default dot; consecutive lines of one marker form one list.
- When the planner outputs contentPlan for add_section, it should be
  directly translatable to edit_page_content arguments:
  sectionType "pricing" and section.content using the keys above. Pricing
  has a single layout — never emit sectionLayout and never use change_layout
  for pricing.

---

# Steps section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock steps section. The styling source of truth
is the steps editing instructions, layout tree, and CSS variable list that are
supplied separately to the planner.

## Layout structure

- Steps has no layout variants: one text column, optionally beside one media
  column (image, GIF, video, or embed) on desktop, media on the left or the
  right. Steps are never a horizontal row and never a multi-column grid.
- The text column is an ordered stack. Every part is optional: a header
  (any/all of eyebrow, title, subtitle), extra content blocks, the step
  list, more extra content blocks, and section CTA buttons. The stack is
  left-aligned and stays inside the text column — with a media column no
  part of it can span both columns.
- The step list is ONE stacked vertical list. Each step is a shared
  decoration badge beside any/all of: a small heading, a description, and
  per-step CTA buttons. A step renders only when at least one of those has
  content.
- The list shares ONE decoration chosen at the section level: an
  auto-incrementing number, one shared icon (built-in check, star, arrow,
  filled success circle, or filled destructive circle, or one custom line
  glyph chosen by name from the built-in library), or none. Every step shows
  the same glyph; only the number changes per step.
- A step row has no container of its own and no per-step media or icon slot.
  Only the decoration badge and the CTA buttons carry backgrounds, borders,
  radius, or shadows.
- Extra content blocks are heading, sub-heading, subtitle, paragraph, image,
  or embed, placed before or after the step list.
- The media column is BOUNDED, never full-bleed. It fills its grid column
  inside the padded content band, sits left or right of the text, and on
  desktop its height is capped to the text column beside it — it can never
  be full-height or edge to edge. Media fit is "fill" (crops to the column)
  or "free-ratio" (natural ratio); the slot can be hidden per
  desktop/mobile. On mobile the section stacks, media above or below the
  text.
- The text-to-media split, the column gap, the content width, the gap
  between steps, and the badge geometry are CSS variables.
- Without a media column the text column is centered at a fixed default
  width; its content stays left-aligned.
- Buttons take a text label (free text; an emoji renders, no library icon)
  plus an optional trailing arrow, in primary/secondary/link variants.
- One media column beside a vertical list of short glyph-led rows belongs
  here, not in alternate: steps repeats one shared glyph for every row. A
  repeating list of full-width [media | text] rows is the alternate section;
  a uniform grid of cards is the boxes family.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- STP-LIM-1: the list is one plain vertical stack with no connector element.
  A free-positioned or diagram-style flow (steps on a canvas, angled or
  branching connectors, a drawn timeline rail between badges) is not stock.
- STP-LIM-2: a line filling as you scroll and active-step highlighting are
  not deliverable. Scroll entrance animation (section reveal and a per-step
  stagger) is built in: the user toggles and tunes it in the section's layout
  settings in the editor, and it is set programmatically with the
  --lr-steps-animation-* variables.
- STP-LIM-3: no horizontal row of steps. Side-by-side step cells, connectors
  or arrows between them, and current-step state are not stock; steps are
  always one vertical list.
- STP-LIM-4: the decoration is one shared badge chosen once for the whole
  list, with no per-item override. An icon is the SAME glyph on every step;
  only a number auto-increments. A design where each step has its own
  distinct icon, shape, or illustration is not stock.
- STP-LIM-5: a step row has no themeable container — no background, border,
  radius, or shadow on the step itself. A design where each step sits in its
  own filled or hairline-bordered card is not stock; the nearest stock drops
  the per-step card and keeps badge + text.
- STP-LIM-6: custom decoration icons are stroke-based line glyphs chosen by
  name from the built-in library; the only filled options are the built-in
  success and destructive circle badges. Solid, duotone, branded, uploaded,
  or exact badge icons are not reproducible; use the nearest line glyph from
  search_icons.
- STP-LIM-7: the section header lives in the text column and is
  left-aligned. A centered header, or a full-width header spanning above
  both the list and the media column, is not stock.
- STP-LIM-8: one bounded media column. Two media slots, per-step images,
  full-bleed or full-height media, media behind the text, or text overlaid
  on the media is not stock.

## Styling

For styling effects, use the supplied steps editing instructions, steps
layout tree, and CSS variable list. Do not duplicate the variable reference
here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include:
section background and padding, content width, the text-to-media column
split and gap, the gap between steps, badge shape/size/fill/border, number
and icon color and size, eyebrow pill styling, per-element typography and
max widths, lines above and below the section title, step description text styling, image radius, and button variants
with hover states for section and item CTAs.

## MCP content contract

- For add_section/update_section, steps content must use renderable keys
  only. Supported section content keys are: eyebrow, title, subtitle, items,
  elements, and ctas (the section CTAs after the list). Do not emit
  bottomCtas, overline, kicker, tagline, heading, or icon.
- The section eyebrow is an object: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- Step items use: id, subtitle, and description, plus optional ctas. The
  step's visible heading is the item subtitle field; there is no item-level
  eyebrow, title, iconName, or image key.
- The decoration is section data, not content: stepDecoration is "number",
  "check", "star", "arrow", "success", "destructive", "custom", or "none",
  with stepIconName (an exact iconName from search_icons) when custom. On
  add_section they sit on the section object beside name and content; on an
  existing section they are changed with edit_page_content
  update_section_property, never through content.
- Extra text blocks are content.elements entries: { id, type: "h1" | "h2" |
  "h3" | "subtitle" | "text", content, position: "before" | "after" } —
  placed before or after the step list. Never author image or embed
  elements.
- Media (image/video/embed) is a user-supplied upload. The planner requests
  the asset instead of authoring it; hide an unused slot with the
  imageVisibility property, set fit with the imageFit property, and set the
  side with the mediaPosition property.
- When the planner outputs contentPlan for add_section, it should be
  directly translatable to edit_page_content arguments: sectionType "steps"
  and section.content using the keys above. Steps has a single layout —
  never emit sectionLayout and never use change_layout for steps.

---

# Tabs section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock tabs section. The styling source of truth is
the tabs editing instructions, layout tree, and CSS variable list that are
supplied separately to the planner.

## Layout structure

- The section is one ordered stack. Every part is optional and renders only
  under the condition stated:
  1. Section header — renders when any of eyebrow, title, subtitle, or header
     CTA buttons carries content. It is a stack of those four, in that order.
     It always sits ABOVE the tabs; a styling variable aligns it centered
     (default) or left. No side-column header exists.
  2. Tab strip — renders when there is at least one tab item. One group of
     clickable labels, exactly one active at a time.
  3. Card — the panel the active tab swaps in. One card per tab item; only the
     active tab's card is on screen.
- Three layouts place the strip: top (a row above the card), left, or right (a
  column beside it). Below the large breakpoint every layout collapses to the
  top arrangement with a horizontally scrollable strip.
- One tab label is an ordered stack of eyebrows, sub headings, and text blocks,
  optionally followed by CTA buttons. A label with a sub heading alone reads as
  a pill; a label carrying text or buttons reads as a card. Which parts render
  is decided by which items carry content, not by a layout setting. Buttons
  always sit below the label text; they cannot be placed between text items.
- In the left and right layouts a data toggle decides label detail: expand
  keeps only the sub heading on inactive tabs and reveals the rest (eyebrows,
  text, buttons) on the active one; show-all shows every part on every tab. In
  the top layout every part that carries content renders.
- The card is a free ordered stack of elements: sub headings, text blocks, CTA
  buttons, images or videos, and embeds, in whatever order the author places
  them. Media is one of those elements, not a fixed slot.
- In the top layout a data toggle splits the card into two columns: media
  elements in one, everything else in the other, both keeping stack order. An
  explicit setting picks which side the media column is on. The left and right
  layouts always stack the card.
- The card area is as tall as the tallest tab's card and does not change per
  tab. Media width comes from its column; only a full-width stacked card
  exposes a per-image size choice.
- The card's copy runs at a readable measure from the card's start edge rather
  than the card's full width. The measure, where that column sits in the card,
  and how the text reads inside it are three independent styling changes; card
  media is not bound by the measure.
- Media fit is the standard fill or free-ratio choice.
- Optional auto-advance rotates the tabs every N seconds (2 to 30), with an
  optional moving progress bar on each tab. It pauses on hover and keyboard
  focus, never runs on touch-style devices, and respects reduced motion.
- Buttons take a text label (free text; an emoji renders, no library icon) plus
  an optional trailing arrow, in primary/secondary/link variants.
- Content the reader clicks between belongs here. Content that is all visible
  at once is the boxes family; a sequential process is the steps section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- TABS-LIM-1: exactly one tab is active at a time and one card is on screen.
  Multiple open panels, an accordion that expands in place, or a card that
  stays visible while another opens is not stock; content that is all visible
  at once belongs in the boxes family.
- TABS-LIM-2: the strip is one flat group. Nested tabs, tab groups within a
  card, and a second independent strip in the same section are not stock.
- TABS-LIM-3: the region model is fixed: header, strip, card. There is no
  bottom text or bottom button region below the card, and no content between
  the strip and the card.
- TABS-LIM-4: a tab label holds eyebrows, sub headings, text, and buttons only.
  A per-tab icon, image, video, or embed on the label itself is not stock; a
  short tag or badge is an eyebrow pill.
- TABS-LIM-5: the card holds sub headings, text, buttons, media, and embeds.
  Tables, logo rows, code blocks, callout boxes, nested testimonials, and tag
  lists are not card elements.
- TABS-LIM-6: the split card is two columns with media on one side. A card
  grid, a masonry card, or more than two columns is not stock.
- TABS-LIM-7: switching tabs crossfades the card at a fixed speed (off under
  reduced motion), and in expand mode the revealed label detail animates open.
  Slide or any other transition, and a custom duration or easing, are not
  stock.
- TABS-LIM-8: auto-advance is a fixed interval with an optional progress bar.
  Per-tab durations, pause/play controls, and prev/next arrows are not stock.
- TABS-LIM-9: the header is always above the tabs, centered or left-aligned. A
  header in a side column beside the strip and card is not stock.

## Styling

For styling effects, use the supplied tabs editing instructions, tabs layout
tree, and CSS variable list. Do not duplicate the variable reference here.

A styling fact is achievable only when it can be mapped to an existing
slot, toggle, or CSS variable from those inputs. Examples to observe
include: section background and padding, content width, the strip/card
column split, the gap between the strip and the card and between tabs,
the strip's vertical alignment against the card in the side layouts, the
spacing between the parts of a tab label and between the parts of the
card, tab surface and border per side, tab radius and padding, the
rest/hover/active state colors, the active indicator line, the progress
bar colors and placement, tab label and card typography, lines above and
below the section title, the card
surface and its split column widths, the card copy's measure and its
alignment within the card, eyebrow pill styling, and header and card
button variants with hover states.

## MCP content contract

- For add_section/update_section, tabs content must use renderable keys only.
  Supported section content keys are: eyebrow, title, subtitle, ctas (the
  header buttons), and items. Do not emit overline, kicker, tagline, heading,
  bottomCtas, or elements.
- The section eyebrow is an object: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- A tab item is { id, labelItems, cardItems } only. title, description, and
  eyebrow on a tab item are rejected by update_section and add_section.
- labelItems is an ordered array holding the tab label's copy. Each entry has id
  and type, where type is one of eyebrow, h3, text, or cta, plus the field that
  type carries: eyebrows (an array of the section eyebrow object) for eyebrow,
  content for h3 and text, ctas for cta. A singular eyebrow field on a label
  entry is rejected. CTA entries are kept last.
- cardItems is an ordered array. Each entry has id and type, where type is one
  of h3, text, cta, image, or embed, plus the field that type carries: content
  for h3 and text, ctas for cta, media/image for image, embed for embed. An
  entry whose required field is missing is dropped from either array.
- The layout is section data, not content: layout is "top", "left", or "right"
  and sits on the section object beside name and content on add_section; on an
  existing section it is changed with edit_page_content change_layout.
- The card and auto-advance settings are section data too, changed with
  edit_page_content update_section_property: tabsCardStyle ("stacked" or
  "split", top layout), tabsMediaSide ("left" or "right", top layout with the
  split card), tabsContentMode ("expand" or "show-all", left and right
  layouts), tabsAutoplay (boolean), tabsAutoplayDuration (2 to 30 seconds), and
  tabsAutoplayProgress (boolean). Media fit uses imageFit.
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "tabs",
  sectionLayout one of top/left/right, and section.content using the keys
  above.

---

# Comparison section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock comparison section. The styling source of
truth is the comparison editing instructions, layout tree, and CSS variable
list that are supplied separately to the planner.

## Layout structure

- The comparison section is a two-column before/after (or us-vs-them)
  contrast: exactly TWO sides in one equal-width grid, stacked on mobile. The
  first side is "before", the second "after".
- There are three comparison layouts:
  - boxes: one image at the top of each side
  - boxes-text: text-only sides, no media
  - boxes-icons: one built-in line icon at the top of each side
- Every comparison section has one optional header and one two-side grid.
- The section header is a stack of any/all of: eyebrow, title, subtitle, and
  CTA buttons. It sits above the sides, centered by default, and can be
  left-aligned while staying above the sides. Unlike boxes, it cannot move to
  a side column. There are no bottom CTAs below the grid.
- Each side is an ordered stack of content rows. Rows can be added, removed,
  reordered, and repeated freely: eyebrow, h3 (the side title), paragraph (a
  plain text block), text (an icon-led bullet row), and cta (a group of
  buttons).
- Each text row can carry one of five built-in glyphs: check, star, arrow,
  success, or destructive — or none. The glyphs are fixed shapes,
  individually recolorable by type. Rows without a glyph indent to keep the
  text edge aligned when sibling rows have glyphs.
- Rows align across the two columns automatically when both sides share the
  same ordered row-type structure; a side whose structure differs flows
  freely instead. To make rows line up, plan matching row order and count on
  both sides.
- In boxes, the side image sits at the top of the side and crops to a square
  by default; the natural aspect ratio is a data toggle. In boxes-icons, the
  icon is chosen by name from the built-in line icon library. A side cannot
  have both an image and an icon.
- Both sides share one width (a single grid max-width) and one gap between
  the columns. Column count is fixed at two.
- There is no dividers toggle and no carousel for comparison; those belong to
  the boxes layouts.
- A uniform grid of three or more cards is the boxes family, not comparison.
  A repeating list of full-width [media | text] rows is the alternate
  section; a shared-glyph vertical list beside media is the steps section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- G-LIM-1: the section has one header and one content region. Side labels,
  split headers, and separate panels between the header and the two columns
  are not part of the comparison layout.
- G-LIM-2: the header is a stack above the columns. A side-column header or
  arbitrary header placement is not supported.
- G-LIM-3: the section has one background across the whole band. Sides carry
  their own card backgrounds/borders, but the grid region has no separate
  background panel, and there is no stock separator rule between the header
  and the columns.
- G-LIM-4: geometry follows the stock section flow. Spacing, widths, and
  alignment can change through styling, but extra layout primitives such as
  absolute-positioned decoration, overlap, or independent panels are not
  stock.
- CMP-LIM-1: exactly TWO sides (before and after). A single column, three or
  more columns, or a full feature matrix is not stock; the nearest stock for
  a grid of three or more cards is the boxes family.
- CMP-LIM-2: the two sides sit side by side in one grid (stacked on mobile).
  No overlap, rotation, z-index layering, or connective arrows, ticks, or
  "VS" badges drawn between the columns.
- CMP-LIM-3: the sides are equal width and cannot be sized independently;
  both are driven by one grid max width. A deliberately wider "after" column
  is not stock.
- CMP-LIM-4: both sides share one structure and one style-variable set.
  Per-side overrides by position can restyle existing slots (for example a
  highlighted "after" card), but not create bespoke markup per column.
- CMP-LIM-5: rows align across the columns only when both sides share the
  same ordered row structure. A true shared table — row separators spanning
  both columns, a feature-label column, or cells merged across columns — is
  not stock.
- CMP-LIM-6: text-row glyphs come from the five built-in shapes (check, star,
  arrow, success, destructive), recolorable by type. A custom glyph, a
  library icon per row, a branded mark, or an uploaded image inside a row is
  not stock; the fallback is the nearest of the five.
- CMP-LIM-7: the boxes-icons side icon comes from the built-in stroke icon
  library. Solid/filled, duotone, branded, uploaded, or exact badge icons are
  not reproducible; use the nearest line glyph from search_icons.
- CMP-LIM-8: each side has at most one media/icon slot, at the top. A side
  needing an image plus an icon, multiple images, media below the text, or a
  full-bleed photo with text overlaid is not stock.
- CMP-LIM-9: no parallax beyond background-attachment. Scroll entrance
  animation (section reveal and item appear) is built in: the user toggles
  and tunes it in the section's layout settings in the editor, and it is set
  programmatically with the --lr-comparison-animation-* variables.

## Styling

For styling effects, use the supplied comparison editing instructions,
comparison layout tree, and CSS variable list. Do not duplicate the variable
reference here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include:
section background and clip-path edges, section padding, header alignment and
max width, the header-to-columns gap, the grid max width and the gap between
the columns, side card backgrounds, borders, shadows, radius, content
padding, content justification, top-image radius and border, top-icon color,
section and side typography, eyebrow pills, the five text-row glyph colors by
type, row stack order, lines above and below the section title and side
titles, CTA styling (section and side), and per-side styling
by position.

## MCP content contract

- For add_section/update_section, comparison content must use renderable keys
  only. Supported section content keys are: eyebrow, title, subtitle, ctas,
  and items. There is no bottomCtas. Do not emit overline, kicker, tagline,
  heading, or icon.
- Section and row eyebrows are objects: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- items holds exactly two sides, in order: { id, type: "before" }, then
  { id, type: "after" }. Side keys are: id, type, image, imageLink,
  openInModal, iconName, and contentItems. For boxes-icons, use iconName from
  search_icons; never use icon for the side icon.
- contentItems is the ordered row list per side. Row shapes:
  { type: "eyebrow", eyebrow: { text, variant } };
  { type: "h3" | "paragraph", content: string };
  { type: "text", content: string, icon: "none" | "check" | "star" | "arrow"
  | "success" | "destructive" };
  { type: "cta", ctas: [...] }. Each row also carries its id.
- Set icon explicitly on every text row; a row without an icon renders no
  glyph. The conventional defaults are destructive rows on the before side
  and success rows on the after side.
- Media (the side image) is a user-supplied upload. The planner requests the
  asset instead of authoring it; image fit is set with the imageFit property
  ("fill" | "free-ratio", boxes layout only).
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "comparison",
  sectionLayout "boxes" | "boxes-text" | "boxes-icons", and section.content
  using the keys above.
- The layout of an existing comparison section is changed with
  edit_page_styles change_layout, not through content.

---

# FAQ section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock FAQ section. The styling source of truth is
the FAQ editing instructions, layout tree, and CSS variable list that are
supplied separately to the planner.

## Layout structure

- FAQ has no layout variants: one optional section header, one vertical list
  of question/answer items, and optional bottom CTA buttons after the last
  item.
- The section header is a stack of any/all of: eyebrow, title, subtitle, and
  CTA buttons. By default it sits ABOVE the questions and is centered; a
  styling change left-aligns it there. A data toggle moves the whole header
  into a LEFT side column beside the questions, left-aligned; the column
  split is a styling change. There is no right column.
- Each item is one uniform row: a question and a collapsible answer, rendered
  as a native browser accordion. Items open and close independently, every
  item starts collapsed, and there is no open-by-default state and no
  single-open behaviour. Scroll entrance animation (section reveal and item
  appear) is built in: the user toggles and tunes it in the section's layout
  settings in the editor, and it is set programmatically with the
  --lr-faq-animation-* variables.
- The question row shows the question text and one built-in plus/minus glyph
  that switches with the open state. Only the glyph's color and stroke width
  are themeable; the glyph itself cannot be swapped.
- The answer is rich text: inline links and bold are stock content with their
  own styling variables. Nothing else can go inside an answer — no media,
  columns, buttons, or bespoke markup.
- The items are one single-column stack; the gap between rows is a styling
  change. Every item shares one card treatment (background, border on all or
  chosen sides, radius, shadow) through styling, or stays plain.
- The bottom region holds CTA buttons only; their alignment is a styling
  change. There is no bottom text slot.
- There is no media slot anywhere in the section and no per-item icon,
  eyebrow, number, or button slot.
- Buttons take a text label (free text; an emoji renders, no library icon)
  plus an optional trailing arrow, in primary/secondary/link variants.
- A question list that expands in place belongs here. A grid of always-open
  Q&A cards is the boxes family; a repeating list of full-width [media | text]
  rows is the alternate section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- FAQ-LIM-1: every item is one uniform row of question plus rich-text answer.
  Inline links and bold are stock; embedded media, columns, buttons, or any
  other bespoke markup inside an answer is not.
- FAQ-LIM-2: the accordion is the native browser element. Open/close
  transitions, open-by-default items, and single-open behaviour (closing one
  item when another opens) are not deliverable.
- FAQ-LIM-3: the region model is fixed: header, question list, optional
  bottom CTA buttons. New regions, a reordered structure, or content between
  questions is not stock.
- FAQ-LIM-4: the expand/collapse icon is the built-in plus/minus glyph; only
  its color and stroke width are themeable. A chevron, arrow, custom glyph,
  or uploaded icon is not stock; the nearest stock keeps the plus/minus glyph
  in the design's color.
- FAQ-LIM-5: the bottom region holds CTA buttons only — there is no bottom
  text slot. A line of text or a prompt beneath the questions is not a FAQ
  element; the nearest stock puts it in a separate section (for example a cta
  section), not custom_design.
- FAQ-LIM-6: the question list is one single column. Two-column or masonry
  question grids are not stock.
- FAQ-LIM-7: no media slot and no per-item icon, number, or image. A design
  with an illustration beside the questions is not stock; the nearest stock
  moves the header to the side column and leaves the illustration out.

## Styling

For styling effects, use the supplied FAQ editing instructions, FAQ layout
tree, and CSS variable list. Do not duplicate the variable reference here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include:
section background and padding, content width, the question-list width, the
gap between the section header and the questions, the left-header column
split, the header's alignment above the questions, item card
background/border/radius/shadow, a border on chosen sides only (for example a
line under each question), the gap between questions, question and answer
typography, lines above and below the section title and question titles,
answer inline link and bold styling, the icon color and stroke
width, eyebrow pill styling, header and bottom button variants with hover
states, and bottom button alignment.

## MCP content contract

- For add_section/update_section, FAQ content must use renderable keys only.
  Supported section content keys are: eyebrow, title, subtitle, ctas (the
  header buttons), items, and bottomCtas (the buttons after the last item).
  Do not emit overline, kicker, tagline, heading, icon, or elements.
- The section eyebrow is an object: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- FAQ items use: id, title, and description. The question is the item title
  field; the answer is the item description field. The answer may carry
  simple inline HTML links and bold; there is no item-level eyebrow,
  subtitle, iconName, image, or ctas key.
- The header placement is section data, not content: sectionHeaderPosition is
  "above" or "left". On add_section it sits on the section object beside name
  and content; on an existing section it is changed with edit_page_content
  update_section_property, never through content.
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "faq" and
  section.content using the keys above. FAQ has a single layout — never emit
  sectionLayout and never use change_layout for faq.

---

# CTA section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock CTA section. The styling source of truth is
the CTA editing instructions, layout tree, and CSS variable list that are
supplied separately to the planner.

## Layout structure

- There are two CTA layouts:
  - default: a centered stack of content elements, no media slot
  - embed: two columns, the content stack left-aligned beside one media/embed
    slot on the right
- The content is an ordered stack of content elements. Elements can be added,
  removed, reordered, and repeated freely: eyebrow, h2 (the CTA title), h3,
  text (subtitle/paragraph), cta (a group of buttons), smallprint, boxes, and
  testimonial. There is no h1 in a CTA.
- boxes is a nested group of 2-4 sub-cards, each with icon, eyebrow, title,
  description, and CTAs. Column counts and card styling are CSS variables.
- testimonial is one quote with author name/role/organization, avatar, media,
  and a logo row.
- The embed layout's media slot takes an image, GIF, video, or embed. A video
  autoplays muted and loops by default (or shows a poster still, or opens in a
  lightbox); a GIF is an animated image; an embed is arbitrary code, so a live
  signup form, calendar, or interactive widget can fill the slot. None of
  these need custom design. The slot can be hidden independently per
  desktop/mobile; fit is "fill" (crops to box) or "free-ratio" (natural
  ratio).
- In the embed layout the media is a bounded side column: it fills its grid
  column inside the content band, is sized by its own aspect ratio, and sits
  vertically centered beside the text. The text-to-media split and the content
  width are CSS variables; the gap between the two columns is fixed.
- Buttons take a text label (free text; an emoji renders, no library icon)
  plus an optional trailing arrow, in primary/secondary/link variants. A
  button links to a URL or anchor, renders embed code inline or as a popup, or
  shows an inline search field.
- A closing conversion band belongs here. A repeating list of full-width
  [media | text] rows is the alternate section; a uniform grid of cards is the
  boxes family; a shared-glyph vertical list beside media is the steps
  section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- CTA-LIM-1: the element kinds are fixed (eyebrow, h2, h3, text, cta,
  smallprint, boxes, testimonial). A content region outside these kinds, or
  bespoke markup inside one, is not stock.
- CTA-LIM-2: real interactive content (live signup forms, calendars, iframes,
  custom widgets) IS stock through the embed layout's embed slot — resolve it
  there, not with custom_design.
- CTA-LIM-3: one media/embed slot, in the embed layout only. The default
  layout has no media slot; two side-by-side media are not stock.
- CTA-LIM-4: the embed layout has no column-gap variable between the text and
  media columns; the space between them cannot be changed.
- CTA-LIM-5: content elements always stack vertically — never side by side
  (heading left, button right). The only side-by-side split is the embed
  layout's text and media columns.
- CTA-LIM-6: no free or absolute positioning of foreground elements; the
  content is a vertical stack. An overlapping foreground collage is not
  stock. Decorative shapes BEHIND content ARE stock — use the background
  layers.
- CTA-LIM-7: media never fills the section edge to edge or floor to ceiling.
  It sits inset in the padded content band — held off the top/bottom by the
  section padding and off the sides by the band's gutter — and that inset
  cannot be zeroed away. A full-bleed or full-height media half is not stock.
- CTA-LIM-8: CTA boxes icons are stroke-based line glyphs chosen by name from
  the built-in library. Solid/filled, branded, uploaded, or exact badge icons
  are not reproducible; use the nearest line glyph from search_icons.
- CTA-LIM-9: the inner content card behind the elements takes a solid color
  or gradient only, never an uploaded image, and the full-section background
  layers are rectangular. A rounded card showing a photo is not stock.
- CTA-LIM-10: a CTA button carries a text label plus an optional trailing
  arrow only. A library icon, inline glyph, or uploaded mark inside a button
  is not stock; an emoji in the label renders, otherwise the nearest stock is
  a text-only button.
- CTA-LIM-11: no inline logo inside a text line (a "by [brand]" byline
  carrying the brand's logo). The eyebrow is text; the nearest stock is
  rendering the brand name as text.
- CTA-LIM-12: no parallax beyond background-attachment, and the CTA has no
  items to stagger. Section reveal on scroll is built in: the user toggles and
  tunes it in the section's layout settings in the editor, and it is set
  programmatically with the --lr-cta-animation-* variables.

## Styling

For styling effects, use the supplied CTA editing instructions, CTA layout
tree, and CSS variable list. Do not duplicate the variable reference here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include: five
stackable background layers (color, image, position, size, repeat, opacity,
z-index, each with mobile overrides), clip-path section edges, the inner
content card (background, padding, radius, shadow, border), per-element
typography, gradient headline and subtitle text, lines above and below the
headline, sub heading and CTA box titles, eyebrow pill styling, button
variants with hover states, adjacent-pair spacing and section padding, the
title-area and subtitle max widths, content width, the embed text-to-media
split and media max width, image radius and border, and CTA boxes and
testimonial card styling.

## MCP content contract

- CTA copy lives in content.elements, an ordered array. When elements is
  non-empty the renderer reads ONLY elements; top-level content.title,
  subtitle, ctas, and smallprint are legacy mirrors — never write them
  alongside elements.
- Element shapes: { type: "eyebrow", eyebrows: [{ text, variant }] };
  { type: "h2" | "h3" | "text" | "smallprint", content: string };
  { type: "cta", ctas: [...] }; { type: "boxes", items: [...] };
  { type: "testimonial", testimonial: { quote, author: { name, position,
  organization } } }. The CTA title is the h2 element; there is no h1.
- Eyebrows are objects: { text: string, variant: "plain" | "pill" } inside the
  element's eyebrows array. A bare eyebrow string, or a singular eyebrow key,
  is not the MCP content shape.
- CTA boxes items use: id, eyebrow, title, description, ctas, and iconName.
  Use iconName from search_icons; never use icon. Do not use iconPosition;
  icon placement is controlled by the CTA boxes icon layout CSS variables.
- Media (image/video/embed) exists only in the embed layout and is a
  user-supplied upload. The planner requests the asset or embed code instead
  of authoring it; hide an unused slot with the imageVisibility property and
  set fit with the imageFit property (embed layout only).
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "cta",
  sectionLayout "default" | "embed", and section.content.elements using the
  shapes above.
- The CTA layout of an existing section is changed with edit_page_styles
  change_layout, not through content.

---

# Custom section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock custom section. The styling source of truth
is the custom editing instructions, layout tree, and CSS variable list that
are supplied separately to the planner.

## Layout structure

- There are two custom layouts:
  - left: one left-aligned vertical column
  - middle: the same column with the text centered
- The section is ONE vertical column of content items stacked top to bottom.
  Items can be added, removed, reordered, and repeated freely: eyebrow, h1,
  h2, h3, text, smallprint, blockquote, code, cta (a group of buttons),
  image, embed, author-date, tags, table, highlight, and logos.
- text is rich text: paragraphs plus inline links and bold, and bulleted or
  numbered lists. The list marker is chosen per list: the default dot, check,
  star, arrow, filled success circle, filled destructive circle, numbers, or
  one line glyph chosen by name from the built-in library. smallprint is a
  fine-print line with the same inline links and bold.
- An eyebrow item holds one or more eyebrows. Several eyebrows in one item
  render as one wrapping row; with the pill variant that row is a free-text
  chip cluster — build chip rows here, not with tags.
- blockquote is a pull-quote: an optional side rule, an optional box
  treatment, and opt-in typographic quote marks per item. It carries inline
  links and bold.
- code is a syntax-highlighted code block (markup, css, javascript,
  typescript, json, bash, or plain) with a built-in hover copy button.
- image takes an image, GIF, or mp4 video (autoplay muted loop, poster
  still, or lightbox); embed is arbitrary code, so a live form, calendar,
  animated or interactive widget is a stock embed item. None of these need
  custom design.
- image and embed items take one of three widths: 600 stays inside the text
  column; 984 breaks out beyond the text column on desktop; 1440 is
  full-bleed edge to edge. There are no other widths.
- author-date is a byline of author name and/or date, in either order. The
  author span carries inline links, bold, and italic.
- tags renders the PAGE's own tags (its tagIds and the workspace tag
  settings), optionally behind a label — never arbitrary text.
- table is a data/comparison grid: one header row plus body rows, a left
  label column, per-column width, alignment, and background, optional row
  striping, an optional line icon per cell chosen by name from the built-in
  library, and inline links and bold in cells. Wide tables scroll
  horizontally.
- highlight is a bordered callout box with its own background and an
  optional media slot beside (or above) its inner stack of eyebrow, h2, h3,
  text, smallprint, cta, and image items. It is the only media-beside-text
  primitive in the section.
- logos is a strip of uploaded logo images, static or auto-scrolling.
- Buttons take a text label (free text; an emoji renders, no library icon)
  plus an optional trailing arrow, in primary/secondary/link variants. A
  button links to a URL or anchor, renders embed code inline or as a popup,
  or shows an inline search field.
- Items are capped by the section content width; text-like items by the
  narrower text column width. Both widths are styling changes.
- An article-like stack of prose, media, callouts, quotes, and tables
  belongs here. A repeating list of full-width [media | text] rows is the
  alternate section; a uniform grid of cards is the boxes family; a question
  list that expands in place is the faq section; a closing conversion band
  is the cta section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- CUS-LIM-1: one vertical column. Items stack top to bottom — no side-by-side
  text columns, no grid, masonry, or bento arrangement, and no text wrapping
  around media. The only multi-region primitives are the table grid and the
  highlight media-beside-content box.
- CUS-LIM-2: the item kinds are fixed (the sixteen listed in Layout
  structure). Real interactive or animated content (forms, calculators,
  widgets, iframes, scripts) IS stock through an embed item — resolve it
  there, not with custom_design. custom_design is only for a section layout
  the single column cannot arrange.
- CUS-LIM-3: image and embed widths are 600, 984, or 1440 only, and every
  other item is capped by the content/text column widths. Free per-item pixel
  widths, absolute positioning, overlap, and z-index layering are not stock.
- CUS-LIM-4: the native items render statically — no tabs, accordions,
  carousels, or reveal-on-click among them, and no parallax beyond
  background-attachment. Interactive behaviour is an embed item, not a native
  one. Section reveal on scroll and a scroll-linked word reveal are built in:
  the user toggles and tunes them in the section's layout settings in the
  editor, and they are set programmatically with the --lr-custom-animation-*
  variables; word reveal is an on/off setting that sweeps the section's whole
  text as one sequence.
- CUS-LIM-5: items are styled per KIND, not per item — every text item shares
  one text style, every h2 one heading style, and so on. Two items of the
  same kind cannot differ in color, size, or weight; the nearest stock for
  two differently styled paragraphs is two stacked custom sections. Table
  and highlight items are the exception: they take per-item styling
  overrides.
- CUS-LIM-6: the column alignment is one choice for the whole section — left
  or centered. Mixing centered and left-aligned items in one section is not
  stock (the highlight box's inner alignment is the one exception); the
  nearest stock is two stacked custom sections.
- CUS-LIM-7: code syntax token colors are a fixed built-in palette. The code
  block's base text, background, border, shadow, radius, and font are
  themeable; a bespoke syntax color theme is not stock.
- CUS-LIM-8: list markers come from the built-in set or the built-in line
  icon library, chosen by name from search_icons. Solid/filled, branded, or
  uploaded marker art is not reproducible; use the nearest line glyph.
- CUS-LIM-9: tags renders the page's own tags only. Arbitrary chip text in a
  tags item is not stock; the nearest stock chip row is an eyebrow item with
  pill-variant eyebrows.
- CUS-LIM-10: a custom CTA button carries a text label plus an optional
  trailing arrow only. A library icon, inline glyph, or uploaded mark inside
  a button is not stock; an emoji in the label renders, otherwise the
  nearest stock is a text-only button.
- CUS-LIM-11: an image has alt text but no caption slot, no themeable
  radius, and no border. The nearest stock caption is a smallprint item
  directly after the image.
- CUS-LIM-12: the section background is one color plus ONE layered
  image/gradient and a clip-path edge — not the stackable multi-layer
  background of other sections. A composition of several decorative
  background shapes is not stock; the nearest stock is one gradient or one
  image layer.

## Styling

For styling effects, use the supplied custom editing instructions, custom
layout tree, and CSS variable list. Do not duplicate the variable reference
here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include:
section background (color plus one image/gradient layer with position, size,
repeat, attachment, origin), clip-path section edges, section padding, the
section content width and the narrower text column width, per-kind
typography and max width for h1/h2/h3/text/eyebrow/smallprint, lines above
and below h1, h2 and h3, inline link
and bold styling, per-kind bottom margins, eyebrow pill styling and the
eyebrow row gaps, blockquote rule/box/quote-mark styling, the code block
container, highlight box background/border/radius/shadow, its media side and
size, and its inner alignment, table borders, striping, header and body
typography, the label column, per-column background and alignment, tag pill
styling, author and date styling, and button variants with hover states.

## MCP content contract

- Custom copy lives in content.items, an ordered array of typed items. There
  are no top-level content keys: do not emit title, subtitle, eyebrow, ctas,
  elements, or bottomCtas — every block is an items entry.
- Item shapes: text, h1, h2, h3, smallprint, and blockquote items use
  content (rich HTML; text carries inline links, bold, and ul/ol lists).
  code items use content as verbatim plain text plus codeLanguage ("markup"
  | "css" | "javascript" | "typescript" | "json" | "bash" | "plain").
  eyebrow items use eyebrows: [{ text, variant: "plain" | "pill" }] — never
  a bare string and never the singular eyebrow key. cta items use ctas.
  embed items use embed.code, never content. blockquote items take an
  optional boolean quoteMarks. author-date items use authorDateOptions
  ({ showAuthor, showDate, order }). image and embed items take imageWidth /
  embedWidth: 600 | 984 | 1440. table items use table ({ columns, rows } of
  cells with text, optional rich content, and optional iconName). highlight
  items use highlight.items with kinds eyebrow, h2, h3, text, smallprint,
  cta, and image. tags items carry no per-item fields — they render the
  page's tagIds with the workspace tag settings.
- Media (image/GIF/video), logos, and embed code are user-supplied uploads.
  The planner authors the item entry and its width but requests the asset
  instead of inventing image URLs, logo entries, or embed markup.
- For add_section, use sectionType "custom", sectionLayout "left" |
  "middle", and section.content.items with at least one item. Never use
  generateWithAI for custom sections.
- For update_section, patch content.items keyed by the existing item ids
  from get_content. update_section can edit, reorder, and remove items but
  never add one — new blocks require add_section of a new custom section.
  Use _itemsMode "replace" with the full item list when reordering or
  removing.
- Custom has no update_section_property toggles.
- The layout of an existing custom section is changed with edit_page_styles
  change_layout, not through content.

---

# Feed section capabilities

Use this guide to decide whether a user-provided description, mockup, or ready
design can be created with a stock feed section. The styling source of truth
is the feed editing instructions, layout tree, and CSS variable list that are
supplied separately to the planner.

## Layout structure

- A feed section renders a live list of post cards pulled from a linked
  workspace collection: each card shows the post's social-preview image,
  title, description, publish date, and a shared read-more label. Use it for
  latest posts, blog indexes, and resource grids.
- There are three feed layouts:
  - boxes: a uniform grid of post cards with the post image on each card
  - boxes-text: the same grid without images — text-only post cards
  - alternate: full-width rows, one post per row as a two-column
    [media | text] split with the image side alternating per row
- Every feed section has one optional header and one region of post items.
- The section header is a stack of any/all of: eyebrow, title, subtitle, and
  CTA buttons. It sits above the items, centered by default, and can be
  left-aligned while staying above the items. It cannot move to a side
  column.
- Post content is dynamic. Card copy, images, links, and dates come from the
  linked collection's posts at render time; nothing per-post is authored in
  this section. Visible card copy in a design is placeholder for live posts.
- Card composition is fixed and uniform: image (boxes and alternate only),
  title, description, date, read-more label — in that order. The description
  and the date can each be hidden for all cards with a data toggle. The
  read-more label is one shared string for every card and renders only when
  set.
- The number of cards is a data toggle (itemCount, 1-100). Post order comes
  from the collection; posts cannot be reordered or hand-picked here.
- In boxes/boxes-text, grid column count, card gap, and card max width are
  styling changes, and whether the image sits above or below the text is a
  styling change. In alternate, the image side pattern (default alternating,
  all left, all right, or first row right) and the mobile order are styling
  changes; alternate has no column control.
- Image fit is a data toggle: "fill" (crops to a fixed wide ratio) or
  "free-ratio" (natural ratio) — boxes and alternate only.
- Cards can be plain or carry a shared card treatment (background, radius,
  border, shadow) through styling, with per-item overrides by position
  (item-0, item-1) bound to the slot, not to a specific post. Cards have
  hover styling for the card background, title, text, and date.
- The bottom region holds ONE optional "view all" CTA button. There is no
  bottom text slot and no second bottom button.
- The section is linked to a collection in the editor. This surface cannot
  list, link, or switch collections.
- A grid of cards whose copy the user authors and hand-orders is the
  collection boxes family, not a feed. A single authored [media | text]
  block belongs to a hero, steps, or cta section.
- See "Styling" and "MCP content contract" for the authoritative styling and
  payload rules.

## Limits

These are the most common limitations to observe and communicate when found:

- G-LIM-1: the section has one header and one content region. Side labels,
  sub-group labels, split headers, and separate panels between header and
  items are not part of the feed layout.
- G-LIM-2: the header is a stack above the items. A side-column header or
  arbitrary header placement is not supported.
- G-LIM-3: the section has one background across the whole band. Cards have
  their own backgrounds/borders, but the items region has no separate
  background panel, and there is no stock separator rule between header and
  items.
- G-LIM-4: geometry follows the stock section flow. Spacing, widths, columns,
  and alignment can change through styling, but extra layout primitives such
  as absolute-positioned decoration, overlap, or independent panels are not
  stock.
- FEED-LIM-1: post content is dynamic. Per-post copy, images, dates, and
  links come from the linked collection and cannot be edited, extended, or
  hand-placed here; only position-bound styling overrides (item-0, item-1)
  are available per card. A design that needs authored, hand-ordered card
  copy is the collection boxes family, not a feed.
- FEED-LIM-2: card composition is fixed: image, title, description, date,
  read-more. The image-versus-text order (boxes) and the image side
  (alternate) can change; adding elements, splitting a card into columns, or
  overlaying text on the image is not stock. Cards carry no category/tag
  pill and no featured-card treatment. When a design's post grid shows tag
  pills or a featured card, ask whether the list must update live or can be
  curated by hand: live keeps this feed section without those elements (a
  per-slot card background tint is the nearest stock highlight); curated is
  a collection boxes section, where a per-card eyebrow pill is the tag.
  Neither needs custom_design.
- FEED-LIM-3: the three layouts are uniform. Masonry, a hero-post-first
  mosaic, mixed card sizes, and column spans are not stock. Grid columns are
  uniform in boxes/boxes-text; alternate has no column control.
- FEED-LIM-4: the card image is the post's social-preview image. Per-post
  cropping, focal control, or substitute media is not stock; the image is
  styled with radius/border and the fill/free-ratio fit toggle. boxes-text
  has no image at all.
- FEED-LIM-5: there is no filter, sort, category-tab, pagination, or
  load-more UI. Order comes from the collection and count from itemCount;
  interactive feed controls are not deliverable with a stock feed.
- FEED-LIM-6: the read-more label is one shared string. Per-card labels or
  differing per-card buttons are not stock; clearing the label removes the
  button from every card.
- FEED-LIM-7: the collection linkage cannot be set from here. When the feed
  is new or the wrong collection is linked, the user picks the collection in
  the editor; name it as a follow-up.

## Styling

For styling effects, use the supplied feed editing instructions, feed layout
tree, and CSS variable list. Do not duplicate the variable reference here.

A styling fact is achievable only when it can be mapped to an existing slot,
toggle, or CSS variable from those inputs. Examples to observe include:
section background and padding, content width, header alignment and the
header-to-items gap, grid columns and card gap and card max width
(boxes/boxes-text), the row gap and the image side pattern (alternate), card
backgrounds/borders/shadows/radius/padding, image radius and border, the
image-versus-text order, title/description/date typography and colors, lines
above and below the section title and card titles, card hover styling, eyebrow pills, read-more button styling, CTA styling (section
and bottom), and per-item styling by position.

## MCP content contract

- For add_section/update_section, feed content must use renderable keys only.
  Supported section content keys are: eyebrow, title, subtitle, and ctas (the
  header buttons). There is no items key and no bottomCtas key — post items
  are never authored, and the closing button is the singular bottomCta on the
  section object, not in content. Do not emit overline, kicker, tagline,
  heading, icon, or elements.
- The section eyebrow is an object: { text: string, variant: "plain" |
  "pill" }. A bare eyebrow string is not the MCP content shape.
- Feed display configuration is section data, not content: itemCount (integer
  1-100), showDescription (boolean), showTimestamp (boolean), readMoreText
  (string; an empty string hides the read-more button), and bottomCta (one
  CTA button object). On add_section they sit on the section object beside
  name and content; on an existing section they are changed with
  edit_page_content update_section_property, never through content. imageFit
  ("fill" | "free-ratio") is the same kind of toggle, valid for boxes and
  alternate only.
- When the planner outputs contentPlan for add_section, it should be directly
  translatable to edit_page_content arguments: sectionType "feed",
  sectionLayout "boxes" | "boxes-text" | "alternate", and section.content
  using the keys above.
- The layout of an existing feed section is changed with edit_page_styles
  change_layout, not through content.
- The collection linkage (which posts render) is not settable through MCP;
  the user links the collection in the editor.


---

# Appendix B: Brand guidelines skill

---
name: brand-guidelines
description: Use this skill when someone wants to turn their brand into a brand guidelines document that LandingRabbit reads cleanly when they sign up or create a new brand. Works from existing guidelines, a website, slides, social posts, or just a font and a few colours.
---

# LandingRabbit Brand Guidelines Skill

## Purpose

When someone signs up for LandingRabbit or creates a new brand, they paste or upload their brand guidelines. LandingRabbit reads that document and builds a design system from it: the colours, the fonts, how buttons and small labels look, how round the corners are, how much room sections get, and a look for every kind of section a page can have.

This skill helps you write that document so nothing gets lost on the way. LandingRabbit applies what the document says, as written. Whatever the document leaves out gets a tested default. So the best document is short, uses LandingRabbit's own names for things, says each thing once, and only talks about what the brand has actually decided.

The document sets up every section type LandingRabbit has, so a page built later on any of them looks right, whether or not the source material shows that kind of section. The shared values cover a section the source lacks. A section gets its own block only when the brand would treat it differently.

LandingRabbit reads the visual side of a brand. It leaves alone everything that is not about how a page looks, for example tone of voice, copywriting rules, logo clear space, photo direction, animations, and what happens when the page scrolls. Those are edited in the app after sign up and design system creation. Highlighted words are the same: once the brand exists, people edit text in the app and can highlight any word there, so the document says nothing about which words are highlighted or what colour they take.

## Working from what the user has

People arrive with all kinds of material: a guideline PDF, a website, a slide deck, a few Instagram posts, a screenshot, or just "our colour is green and we use Poppins". All of it works.

When the material is images or slides rather than a written spec, read it for the brand's essence: for example the two or three colours that keep coming back, whether the type is heavy or light, serif or sans, whether corners are soft or sharp, whether the layout feels dense or airy, and whether the brand leans dark or light. Then picture what a typical website for this brand would look like, and write the document from that picture. Pick exact values that match what you see. Say in one line at the top of the document which values you chose yourself, so the user knows what to check.

When the material is a website, its stylesheet is usually the most precise source. Colour variables, font names, button styles and corner radius come straight from there.

Whatever the material is, it is where the values come from, not the list of what to cover. The list of what to cover is the Sections table. A page or block in the source with no counterpart in that table gives its colours and shapes to the shared values and is otherwise left out.

## Step-by-step task

1. Collect what the user has, in any form.
2. Ask only for the two things the document cannot do without: at least one font and the main colours. Everything else comes from the material or from LandingRabbit's defaults.
3. Pick one value per role. When a source gives several, the web version wins over print or slides.
4. Write the shared values first, using the skeleton under "Document skeleton" and the names under "Terminology". Then go through every row of the Sections table, decide whether the brand treats that section differently from the shared values, and write a block for it only when it does.
5. Read the document once from top to bottom for each entry under "The common slips", fix what you find, then reply with the document on its own.

## The common slips

These are the places where a first draft usually goes wrong. Read them before writing and walk through them after, in this order.

- **A section or layout that is not in the table.** Every section and layout name comes from the Sections table. Anything else is renamed to the nearest one there, or left out.
- **An element with no slot.** Every named element is in the Sections table, the navigation and footer rows included. Same treatment.
- **A decision made per piece of content.** Which words are highlighted, which words are bold, which plan is featured, which item comes first. Those are made in the editor once the brand exists, so the document leaves them out.
- **A word from the source.** Every noun is checked against the translation table.
- **A colour value outside Palette.** Every colour in the document is a named entry in Palette. A new colour gets a name there first.
- **A range, a rough figure or two units.** One value, one format, throughout. Where the source gives a range or two alternatives, the document states the one value the brand would choose.
- **A tier larger than the one above it.** Sizes run from 4xl down to xs. A tier that comes out bigger than the tier above it is checked against the source before it stays.
- **A weight or size given to "headings" or "body".** Given to tier names instead, the way the example does it.
- **Something that happens.** Anything on scroll, focus, click or load. Hover on buttons and links stays. The rest goes.
- **A rule that depends on the situation.** Two plain rules on two elements, or none.
- **Two styles for one kind of item.** Every item in a section shares one style, in comparison as much as anywhere.
- **A fact said twice.** The second copy goes. Brand holds the feel in words, the other blocks hold the values, and a section block holds only what differs from the shared values. The usual place this happens is a section block that repeats the item fill, the item border, or a text colour the shared values or the dark-surface paragraph already give.
- **Text around the document.** The reply is the document.

And finally: every palette colour is used by name at least once, every button role has a rest look and a hover look, every background names the text colours that sit on it, every font is on Google Fonts, is a system font, or names a Google Fonts substitute, and the document is under 25,000 characters.

## Rules

- Say each thing once. The shared values apply everywhere. A section block holds only what is different in that section.
- Name each colour once under Palette, with its value. Everywhere else, use the name.
- Use LandingRabbit's names for sections and elements. The translation table turns everyday design words into those names.
- Write rules that hold for everything in their scope. When a look depends on the situation, write it as two plain rules on two elements, or leave it out.
- All items in a section share one style. A comparison is two or more items, for example a before and an after, and they share one style too.
- Hover is the one interaction the design carries, on buttons and links. Give every button role a rest look and a hover look.
- Give every background the text colours that sit on it: title, subtitle, smallprint and eyebrow, and the buttons when they change.
- Use Google Fonts or system fonts. When the brand font is not on Google Fonts, name the closest Google Fonts family after the word "Substitute:". A system font is matched to its closest Google Fonts family, so naming that family directly gives the exact result. Anyone who wants their own font files used can contact the LandingRabbit team after the brand is generated.
- Use one value format throughout: hex and px, or hsla and rem.
- Leave out whatever the brand does not define. LandingRabbit fills it in with a default that works.
- Keep the whole document under 25,000 characters.
- Write in plain, everyday language, the way you would explain the brand to a new colleague. People who are not designers read this document, and they will ask you questions about it.
- Write the document as markdown, with the heading marks and pipe tables the skeleton and the example use.
- Reply with the document and nothing else.

## How LandingRabbit reads the document

- Shared values become the brand: the page background, the text colours, the fonts, the buttons and the eyebrow. Every section inherits them.
- Section blocks become that section's own settings and win over the shared values inside that section.
- Words about shape and labels, for example "sharp corners", "pill buttons" or "labels in small rounded tags", pick the structural starting point the brand is built on.
- A section described the way one of the named looks works gets that look as its default.
- A text role given a size tier reads that tier everywhere it appears.
- Colours are understood by the job they do. "Primary action" and "border" tell LandingRabbit more than "brand blue" and "grey 200".

## Terminology

### Page anatomy

A LandingRabbit page is a vertical stack of sections between a navigation bar and a footer.

- **Page background**: the colour behind everything. A section without its own background shows it through.
- **Section**: one block of the page, of one type from the Sections table. It can have its own **section background**.
- **Section header**: the top part of most sections. It holds a **section eyebrow**, a **section title**, a **section subtitle** and **section buttons**. The header sits above the content or, in some looks, beside it. It is aligned left or centred.
- **Bottom buttons**: buttons under a section's items, at the end of the section.
- **Item**: the repeated unit inside a section, for example one card in a feature grid, one plan in pricing, or one question in the FAQ. An item can have an **item eyebrow**, an **item title**, **item text**, an **item icon** and **item buttons**. It can have a fill, a border, a radius, a shadow and padding.
- **Element**: a single additional card behind the hero content, the CTA content, or a testimonial. It can have a fill, a border, a radius, a shadow and padding.
- **Nested boxes item**: a small feature or stat box inside the hero, the CTA, or an alternate row. It has the same parts as an item.
- **Navigation**: the bar at the top with the logo, links and buttons.
- **Footer**: the block at the bottom with the logo, link groups, text and smallprint.

### Colour roles

LandingRabbit sorts colours by the job they do. Give every colour in the palette at least one of these roles.

- **page background**: the colour of the page itself.
- **primary text**: headings and body copy.
- **muted text**: smallprint, captions, dates and other quiet text.
- **primary action**: the main button.
- **secondary action**: the second button, usually quieter.
- **accent**: highlights, for example the eyebrow, icons and list markers.
- **section surface**: a section background that differs from the page background.
- **item surface**: the fill of items and elements.
- **border**: hairlines, dividers and item borders.
- **navigation**: the navigation bar.
- **footer**: the footer.

One dark surface is enough for a brand. Say where it sits, for example the hero, the CTA, or the footer, and give the text colours that go on it once.

### Typography

Text sizes come in nine tiers. Each tier has a family, a size, a mobile size, a weight, a line height, a letter spacing and a case. State the properties the brand defines and leave the rest.

| Tier | What reads it by default                                                                                                     |
| ---- | ---------------------------------------------------------------------------------------------------------------------------- |
| 4xl  | hero title                                                                                                                   |
| 3xl  | section titles, CTA title                                                                                                    |
| 2xl  | hero h3                                                                                                                      |
| xl   | alternate item title, trustedBy h2, custom h2, nested boxes item titles inside hero and CTA                                  |
| lg   | item titles in boxes, comparison and feed, steps h3, custom h3, trustedBy h3, testimonial quote, pricing price and plan name |
| md   | hero subtitle, CTA subtitle                                                                                                  |
| base | section subtitles, alternate item text, steps text, FAQ and tabs item titles                                                 |
| sm   | item text in boxes, comparison, feed, FAQ, pricing and tabs                                                                  |
| xs   | custom author, date and tags                                                                                                 |

A weight, size, line height, letter spacing or case is always given to tier names, for example "tiers 4xl, 3xl, 2xl and xl: weight 800", so every tier ends up with one clear value.

Eight text roles can be given a tier of their own: **section subtitle**, **item title**, **item text**, **quote**, **alternate item title**, **price**, **testimonial name** and **testimonial meta**, which is the position and organization line. Write a page-wide assignment as "item titles read xl" and a one-section one as "pricing: item titles read lg". A role given a tier can also be given its own weight, letter spacing, line height or case, for example "quotes read 2xl at weight 300". That applies to the role only, not to the tier.

The eyebrow and the buttons have their own text settings, separate from the tiers. Navigation and footer text sizes are stated directly in their blocks.

### Buttons

Three button roles: **primary**, **secondary** and **link**. Each has a **rest** look and a **hover** look. Properties: for example fill, border, text colour, radius, horizontal and vertical padding, font weight, font size, letter spacing, case, and underline on the link. The same three roles appear inside sections, items and the navigation, and inherit these values.

### Eyebrow

The eyebrow is the short label above a title: in a section header, in the hero, in the CTA, or inside an item. It renders **plain** or in a **pill**. Text properties: colour, size, weight, letter spacing, line height, case. Pill properties: fill, border colour, border width, radius, opacity, padding. Note that text and pill (background) colour need to work together

### Shape, borders and shadows

A radius can sit on buttons, the eyebrow pill, items, elements, nested boxes items, the icon container, images, the step decoration (usually icons and numbers), and the tabs list and card. A shadow can sit on items, elements and nested boxes items. A border can sit on items, elements, nested boxes items, images and the icon container. State one value per slot the brand defines. A single radius for the whole brand is a complete answer.

### Icons and markers

- **Item icon**: the icon in a boxes-icons item and in nested boxes items. Properties: colour, size, stroke width, plus its **icon container** fill, radius, padding and minimum size. Layout: icon above the text or beside it.
- **List markers**: bullets in any text. Kinds: check, success, destructive, star, arrow, dot. Each kind has a colour. Size and gap are shared.
- **Step decoration**: the mark before each step in the steps section, a number or an icon. Properties: fill, border, colour, minimum size, radius, and the number's font.

### Layout

- **Content width**: how wide sections are. **Narrow content width** for text-only sections.
- **Section padding**: a roomy value and a tight value, top and bottom.
- **Header gap**: the space between a section header and its content.
- **Header alignment**: left or centred, page-wide, with a per-section exception when one section differs.
- **Hero text width**, **CTA stack width**, **FAQ list width** and **quote width**.
- **Nested boxes columns**: how many nested boxes items sit in a row in the hero and the CTA.

### Sections

These are all the section types a LandingRabbit page can have, plus the navigation and footer around them.

| Section      | Layouts                                                                                                     | Elements                                                                                                                                            | Named looks                                                                     |
| ------------ | ----------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------- |
| hero         | horizontal (text left, media right), horizontal-reversed (media left), vertical (centred text, media below) | eyebrow, title, subtitle, buttons, smallprint, media, element, nested boxes items, testimonial                                                      | Background element (a card behind the text), Large media, Images on left        |
| trustedBy    | one                                                                                                         | eyebrow, h2, h3, text, buttons, logos                                                                                                               |                                                                                 |
| boxes        | boxes (cards with an image), boxes-text (text cards, also stat tiles), boxes-icons (cards with an icon)     | section header, items with item eyebrow, title, text, icon and buttons, columns, divider, bottom buttons                                            | Grid, Top border, List view center, List view left, Header on the side, Divider |
| alternate    | rows that alternate image left and image right                                                              | section header, rows with a box, an image, eyebrow, title, text, buttons, nested boxes items and a testimonial, bottom buttons                      | Boxes Two Columns                                                               |
| testimonials | carousel, grid                                                                                              | section header, element (carousel), grid element, quote, author name, position, organization, avatar, logo, bottom buttons                          |                                                                                 |
| pricing      | one                                                                                                         | section header, items with item eyebrow, plan name, price, price period, item subtitle, item list icon and item button; billing selector and toggle |                                                                                 |
| steps        | one, media left or right                                                                                    | eyebrow, title, subtitle, items with step decoration, h3 and text, media, buttons                                                                   |                                                                                 |
| tabs         | top (tabs above the card), side (tabs beside the card)                                                      | section header, tabs list, tab item with rest, hover and active looks, active line, accent, progress bar, card with title, text, buttons and media  | Pill, Underline, Header left, Side line, Cards with side line                   |
| comparison   | boxes, boxes-text, boxes-icons                                                                              | section header, two or more items sharing one style, for example a before and an after, each with icon, title, text and buttons                     |                                                                                 |
| faq          | one                                                                                                         | section header, items with item title (the question), item text (the answer) and item icon                                                          | Header on the side                                                              |
| cta          | default (centred text and buttons), embed (text beside a video, form or image)                              | eyebrow, title, subtitle, buttons, smallprint, element, nested boxes items, media, testimonial                                                      | Background element, Large media                                                 |
| custom       | left, middle                                                                                                | h1, h2, h3, text, blockquote, code, highlight box, table, tags, author, date, image, embed, buttons                                                 |                                                                                 |
| feed         | boxes, boxes-text, alternate                                                                                | section header, items with title, text, date and read-more button                                                                                   |                                                                                 |
| navigation   |                                                                                                             | background (a solid colour), bottom border, height, logo, links, buttons                                                                            |                                                                                 |
| footer       |                                                                                                             | background (a solid colour), top border, logo, group titles, links, text, smallprint, social icons                                                  |                                                                                 |

The hero and the CTA carry their own background, flat or gradient or image, and an optional element behind the text with its own fill. A gradient is written as a CSS gradient, for example "hero background: linear-gradient(135deg, ink, ink-deep)", followed by the text colours that sit on it.

This table is the whole list. An element in the source with no slot here is described with the nearest element in this table or left out. A section in the source with no counterpart here is left out.

### Translation table

| The source says                             | LandingRabbit says                                                                  |
| ------------------------------------------- | ----------------------------------------------------------------------------------- |
| card, tile, panel, container, module        | item, or element when it is the single card behind hero, CTA or testimonial content |
| ground, canvas, paper, base                 | page background or section background                                               |
| plate, band, strip                          | section background                                                                  |
| nested box, callout, stat tile, feature box | nested boxes item                                                                   |
| chip, tag, badge, label, kicker, overline   | eyebrow                                                                             |
| lead, standfirst, deck, intro               | subtitle                                                                            |
| headline, display, heading                  | title                                                                               |
| body, copy, paragraph                       | text                                                                                |
| knob                                        | toggle thumb                                                                        |
| pill track, segmented control               | tabs list, with the Pill look                                                       |
| hairline, rule, stroke                      | border                                                                              |
| numeral, step number                        | step decoration                                                                     |
| logo wall, client strip, trust bar          | trustedBy                                                                           |
| feature grid, benefits, services, card grid | boxes                                                                               |
| alternating rows, zigzag, feature narrative | alternate                                                                           |
| process, how it works, timeline             | steps                                                                               |
| plans, tiers, packages                      | pricing                                                                             |
| versus, before and after                    | comparison                                                                          |
| accordion, Q&A                              | faq                                                                                 |
| banner, sign-up block                       | cta                                                                                 |
| about, story, article                       | custom                                                                              |
| blog, news, resources                       | feed                                                                                |

## Document skeleton

Use these headings, in this order. Leave out a heading that has nothing to say.

```
# <Brand name> brand guidelines
## Brand
## Palette
## Colour roles
## Typography
## Buttons
## Eyebrow
## Shape, borders and shadows
## Icons and markers
## Layout
## Navigation
## Footer
## Sections
### hero
### cta
### <any other section that differs from the shared values>
```

Brand holds the feel in words, for example calm or bold, editorial or technical, serif or sans, soft or sharp corners, dense or airy, plain or pill labels. The values live in their own blocks: for example case, sizes, colours, and where a surface sits are values.

Hero and cta always appear under Sections because they carry their own backgrounds. Every other section appears only when something in it differs from the shared values.

## Example

A complete document for a fictional brand. The palette is neutral with one accent, and the hero and the CTA sit on a dark gradient. Everything not written here is left to LandingRabbit's defaults. It shows the format only. Choose every value from the person's own brand, not from this example.

```markdown
# Ledgerline brand guidelines

Colours are hsla, lengths are rem at a 16px root. Every fact appears once. Colours are named under Palette and referred to by name after that.

## Brand

Ledgerline is bookkeeping software for small agencies. The look is calm and precise: a white page, near-black text, one teal accent for actions, hairline borders instead of shadows, and a dark gradient band for the big moments. One sans-serif family for headings and another for text, headings heavy, text regular. Corners softly rounded. Labels plain.

## Palette

- ink hsla(220, 15%, 12%, 1)
- ink-deep hsla(220, 20%, 22%, 1)
- slate hsla(220, 10%, 42%, 1)
- white hsla(0, 0%, 100%, 1)
- fog hsla(0, 0%, 100%, 0.8)
- glass hsla(0, 0%, 100%, 0.06)
- mist hsla(220, 14%, 96%, 1)
- line hsla(220, 12%, 90%, 1)
- line-on-ink hsla(0, 0%, 100%, 0.18)
- teal hsla(190, 60%, 32%, 1)
- teal-deep hsla(190, 60%, 26%, 1)
- teal-light hsla(190, 55%, 62%, 1)
- coral hsla(8, 70%, 52%, 1)

## Colour roles

- page background: white
- primary text: ink
- muted text: slate
- primary action: teal
- secondary action: ink, outlined
- accent: teal
- section surface: mist
- item surface: white
- border: line
- navigation: white
- footer: ink

The dark surface is ink. It sits on the hero background, the CTA background and the footer. Everything on ink uses: title white, subtitle fog, smallprint fog, text fog, eyebrow teal-light, border line-on-ink, item icon teal-light. Nested boxes items on ink: glass fill, line-on-ink border, title white, text fog. Secondary button on ink at rest: transparent fill, line-on-ink border, white text; hover: white fill, ink text. Testimonial on ink: quote and author name white, position and organization fog.

## Typography

Tiers 4xl, 3xl, 2xl and xl: Manrope, weight 800, letter spacing -0.02em, line height 1.1.
Tiers lg, md, base, sm and xs: Inter, weight 400, line height 1.5.

| Tier | Size     | Mobile size |
| ---- | -------- | ----------- |
| 4xl  | 3.5rem   | 2.5rem      |
| 3xl  | 2.5rem   | 2rem        |
| 2xl  | 2rem     | 1.75rem     |
| xl   | 1.5rem   | 1.375rem    |
| lg   | 1.25rem  | 1.25rem     |
| md   | 1.25rem  | 1.125rem    |
| base | 1.125rem | 1rem        |
| sm   | 1rem     | 1rem        |
| xs   | 0.875rem | 0.875rem    |

Item titles read xl. Section subtitles are slate.

## Buttons

Radius 0.5rem. Padding 1.5rem horizontal, 0.75rem vertical. Label Inter, 1rem, weight 600.

- primary at rest: teal fill, white text, no border. Hover: teal-deep fill.
- secondary at rest: transparent fill, ink border, ink text. Hover: ink fill, white text.
- link at rest: teal text, no underline. Hover: teal-deep text, underline.

## Eyebrow

Plain. Inter, 0.75rem, weight 600, letter spacing 0.08em, uppercase, teal.

## Shape, borders and shadows

Radius 1rem on items, elements and nested boxes items. Radius 0.5rem on images, the icon container and the step decoration. Every border is 1px wide. Items and elements have a line border. Shadows none.

## Icons and markers

Item icon: teal, 1.5rem, stroke width 1.75, above the text. Icon container: mist fill, 3rem minimum size.
List markers: check, success, star and arrow teal; destructive coral; dot slate.
Step decoration: a number, mist fill, line border, ink number in Inter weight 600, 2.5rem minimum size.

## Layout

Content width 80rem. Narrow content width 48rem. Section padding 6rem roomy, 3rem tight. Header gap 3rem. Section headers left-aligned. Hero text width 40rem. CTA stack width 44rem. FAQ list width 48rem. Nested boxes columns 2.

## Navigation

White background, line bottom border, height 4.5rem. Links ink, Inter 1rem, weight 500; hover teal.

## Footer

Ink background, line-on-ink top border. Group titles white, Inter 0.875rem, weight 600, uppercase, letter spacing 0.08em. Links fog, hover white. Text and smallprint fog. Social icons fog.

## Sections

### hero

Layout horizontal. Background linear-gradient(135deg, ink, ink-deep).

### cta

Layout default. Background linear-gradient(315deg, ink, ink-deep).

### steps

Section background mist.

### tabs

Underline look. Accent teal. Card: white fill, line border.

### faq

Divider look. Divider line.

### testimonials

Section background mist.

### comparison

Section background mist.
```

## Showing what the page will look like

Once the document is done, the next question is usually "Can you show me how a page with this would look?" You will build a mockup. Build it so the person recognises the page when they see the real one in LandingRabbit.

The mockup is made of LandingRabbit sections and nothing else. Every block in it is one section type from the Sections table, in one of its layouts, holding only the elements that table lists for it. The page reads top to bottom as a stack of those sections between a navigation bar and a footer. A page usually has a hero first and a cta near the end, with for example a trustedBy, a boxes, a steps and a faq in between.

Every visual value in the mockup comes from the document. Colours are the palette, fonts and sizes are the typography tiers, buttons and eyebrows are the ones the document describes, corners and borders follow the shape values, and section backgrounds are the ones the document assigns. Whatever the document leaves out renders plainly: no fill, no border, no shadow, text in the primary text colour on the page background.

The gallery at app.landingrabbit.com/design shows every section type, layout and named look rendered with placeholder content, one page per template. Use it as the reference for what a section contains and how its elements sit: for example where the eyebrow goes, how nested boxes items line up inside a hero, what a Top border item looks like, how the step decoration sits beside a step. Take the elements and their arrangement from there. Take the colours, fonts, spacing feel and mood from the document alone. The templates on those pages are other brands' looks, and none of that carries over.

Keep the mockup honest in these ways:

- Items in a section all share one style, comparison items included.
- The only state is hover, on buttons and links.
- Elements that are not in the Sections table stay out, for example ribbons, floating badges, decorative shapes, illustrations, chat bubbles or custom widgets.
- Placeholder text stands in for copy unless the user has given theirs or you know their company.
- One line under the mockup says which values came from the document and which are LandingRabbit defaults, and that the real page is rendered by LandingRabbit from the document, so small differences in spacing and sizing are expected.

When asked to change the mockup, change the document first, then the mockup, so the two stay the same thing.
