launchframe 0.4.4 → 0.4.5

package/.amazonq/cli-agents/launchframe.json

@@ -1,7 +1,7 @@
 {
   "name": "launchframe",
   "description": "Reverse-engineer a reference URL into this repo + SaaS landing copy — /launchframe",
-  "prompt": "\n# Launchframe\n\n**`/launchframe`** is the **only** slash command for reverse-engineering a live site into this codebase. Parse **the reference URL(s) and SaaS idea the user passed with /launchframe**:\n\n1. **Reference URL(s)** — Every `http://` or `https://` token (clone each site; use `docs/research/<hostname>/` when there are multiple).\n2. **SaaS idea** — All remaining text that is not a URL (often quoted): product pitch for hero and key marketing lines.\n\nYou are a **foreman walking the job site** — write specs per section, dispatch builders in parallel where possible, merge, QA. Not a shallow inspect-then-guess flow.\n\n## Step 0 — Persist Launchframe inputs\n\nBefore reconnaissance, write or update:\n\n- `src/lib/launchframe-config.ts` — `LAUNCHFRAME_SOURCE_URL` (primary URL, usually first), `LAUNCHFRAME_SAAS_IDEA`\n- `launchframe.context.json`\n- `docs/research/LAUNCHFRAME.md` — URLs and SaaS idea\n\n## SaaS copy overlay (Phase 4 assembly and final polish)\n\nAfter structure and styles match the reference, apply the **SaaS idea** to hero, headings, and primary CTAs where the reference uses interchangeable marketing copy, **without** changing layout grids, spacing, or motion from extracted specs. **Brand identity** (below) must be **original** for anything you ship as the user’s product — never pass off the reference company’s trademarks or distinctive marks.\n\n**Precedence:** For those surfaces (hero, main headings, primary CTAs), **final ship copy comes from the SaaS idea**, not the reference — still capture reference strings in research/specs if useful for QA diffs. Everywhere else, keep **verbatim** reference text unless the user overrides.\n\n---\n\n## Scope Defaults\n\nThe target page(s) are the URL(s) you parsed in Pre-Flight. Clone exactly what's visible at each URL. Unless the user specifies otherwise, use these defaults:\n\n- **Fidelity level:** Pixel-perfect **layout, spacing, typography scale, and motion** (durations, easings, keyframes, triggers) matched to the reference. **Colors** match where you are emulating neutral/UI chrome from the reference; where **Brand identity** applies, use the **new palette** and assets while keeping the same CSS structure (fills land on different tokens). **Animations** stay reference-faithful unless the user opts out.\n- **In scope:** Visual layout and styling, component structure and interactions, responsive design, mock data for demo purposes\n- **Out of scope:** Real backend / database, authentication, real-time features, SEO optimization, accessibility audit\n- **Customization:** Structure and visuals — pure emulation of the reference. **Marketing copy** — apply the parsed **SaaS idea** where hero/headlines/CTAs are interchangeable (see “SaaS copy overlay” above); do not invent a different product than the user’s idea.\n- **Brand identity:** The reference is a **pattern** for layout, motion, and UI craft — **not** permission to ship their brand. Unless the user **explicitly** asks for a faithful copy of the reference brand (e.g. licensed work, clearly labeled internal mock, private design audit), **invent an original brand** aligned with the SaaS idea: product name, wordmark or simple logomark (SVG or styled text) sized to the same logo slot, favicon / app icon, OG imagery, and a cohesive palette. Do **not** reuse their trademarked logo paths, mascot art, or distinctive illustrative brand assets; use originals or functional UI icons instead. Hero or lifestyle images that center the reference brand should be replaced with **original** imagery or neutral compositions that keep the **same layout rhythm**. **Keep** reference-faithful neutrals where they are clearly **non-brand** UI (borders, subtle grays, default body text color roles) unless they are unmistakably part of their distinctive palette. Note in `docs/research/LAUNCHFRAME.md` which marks and assets are **original brand** versus **layout-only** extraction.\n\nIf the user provides additional instructions (specific fidelity level, customizations, extra context), honor those over the defaults.\n\n## Pre-Flight\n\n1. **Browser automation is required.** Check for available browser MCP tools (Chrome MCP, Playwright MCP, Browserbase MCP, Puppeteer MCP, etc.). Use whichever is available — if multiple exist, **prefer Chrome DevTools MCP** for inspection: it maps to the same engine the user sees and preserves **computed `animation` / `transition` / scroll-driven** values accurately. If none are detected, ask the user which browser tool they have and how to connect it. This skill cannot work without browser automation. **Motion parity:** every pass (reconnaissance, per-section extract, QA) must treat **motion as first-class** — same **durations, delays, easings (`cubic-bezier` / steps), iteration counts, fill modes, directions, keyframe percentages, and scroll/view-timeline bindings** as the live site, not “similar” motion.\n2. **Parse arguments** — extract every `http://` / `https://` URL token (there may be several). **SaaS idea** = the remaining non-URL text (trim outer quotes). Normalize and validate each URL; if any are invalid, or the SaaS idea is missing, ask the user once. For each valid URL, verify it is accessible via your browser MCP tool.\n3. Verify the base project builds: `npm run build`. The Next.js + shadcn/ui + Tailwind v4 scaffold should already be in place. If not, tell the user to set it up first.\n4. Create the output directories if they don't exist: `docs/research/`, `docs/research/components/`, `docs/design-references/`, `scripts/`. For multiple clones, also prepare per-site folders like `docs/research/<hostname>/` and `docs/design-references/<hostname>/`.\n5. When working with multiple sites in one command, optionally confirm whether to run them in parallel (recommended, if resources allow) or sequentially to avoid overload.\n\n## Guiding Principles\n\nThese are the truths that separate a successful clone from a \"close enough\" mess. Internalize them — they should inform every decision you make.\n\n### 0. Visual crawl priority (images, SVGs, motion — first)\n\nWhen you traverse the DOM and the Network panel, do **not** treat all nodes equally. Work in this **priority order** so the clone feels like the original, not a wireframe:\n\n1. **Images (raster + video stills)** — Enumerate `<img>`, `<picture>`, responsive `srcset`, `data-*` lazy URLs, **computed `background-image`** on the element and parents (including pseudo-elements), mask images, `<video poster>`, hero media. **Scrape:** download binary assets to `public/images/` (or `public/videos/`) with stable paths referenced in specs. **Create** a replacement PNG/WebP/SVG only when the asset is blocked (CORS, auth cookie, 403), ephemeral, or impossible to URL-fetch — use a high-DPI screenshot crop, traced artwork, or CSS gradient approximation, and mark `ASSET_SOURCE: generated` in the component spec with a short reason.\n2. **SVGs & iconography** — Inline `<svg>`, sprite `symbol` defs, **SVG used as masks/filters**, icon fonts (prefer path extraction). Convert to `@/components/icons.tsx` (or section-local components) with meaningful names. Prioritize crisp edges and correct `viewBox` over shrinking bundle size during emulation.\n3. **Motion & animation** — CSS `@keyframes`, `animation`, `animation-timeline` / `view-timeline` / `animation-range`, `transition`, `transform`, `transform-origin`, will-change; JS-driven motion (carousel timing, IntersectionObserver reveals); libraries (GSAP, Framer, Lottie JSON, Lenis). Capture **numbers** (ms, `cubic-bezier()`, `steps()`, stagger, scroll thresholds, **named keyframe blocks**), not adjectives — paste **verbatim** `cssText` for relevant `@keyframes` into specs when obtainable. Include **reduced-motion** (`matchMedia('(prefers-reduced-motion: reduce)'))` behavior if present.\n\nOnly after the above are accounted for should you spend cycle time on minor text or non-visual refactors. A perfect grid with missing hero art and dead animation still fails the clone.\n\n### 1. Completeness Beats Speed\n\nEvery builder agent must receive **everything** it needs to do its job perfectly: screenshot, exact CSS values, downloaded assets with local paths, real text content, component structure. If a builder has to guess anything — a color, a font size, a padding value — you have failed at extraction. Take the extra minute to extract one more property rather than shipping an incomplete brief.\n\n### 2. Small Tasks, Perfect Results\n\nWhen an agent gets \"build the entire features section,\" it glosses over details — it approximates spacing, guesses font sizes, and produces something \"close enough\" but clearly wrong. When it gets a single focused component with exact CSS values, it nails it every time.\n\nLook at each section and judge its complexity. A simple banner with a heading and a button? One agent. A complex section with 3 different card variants, each with unique hover states and internal layouts? One agent per card variant plus one for the section wrapper. When in doubt, make it smaller.\n\n**Complexity budget rule:** If a builder prompt exceeds ~150 lines of spec content, the section is too complex for one agent. Break it into smaller pieces. This is a mechanical check — don't override it with \"but it's all related.\"\n\n### 3. Real Content, Real Assets\n\nExtract the actual text, images, videos, and SVGs from the live site. This is a clone, not a mockup. Use `element.textContent`, download every `<img>` and `<video>`, extract inline `<svg>` elements as React components. The only time you generate content is when something is clearly server-generated and unique per session.\n\n**Marketing surfaces** (hero, primary headings, main CTAs) follow the **SaaS copy overlay** and **Brand identity** rules — capture reference copy for audit, but **ship** the user’s idea there.\n\n**Prioritize** (see §0): downloadable imagery and backgrounds first, then SVG/icon layers, then motion. If you must **fabricate** an asset, prefer screenshot-based exports or traced vectors tied to measured box sizes — avoid unrelated stock art.\n\n**Layered assets matter.** A section that looks like one image is often multiple layers — a background watercolor/gradient, a foreground UI mockup PNG, an overlay icon. Inspect each container's full DOM tree and enumerate ALL `<img>` elements and background images within it, including absolutely-positioned overlays. Missing an overlay image makes the clone look empty even if the background is correct.\n\n### 4. Foundation First\n\nNothing can be built until the foundation exists: global CSS with the target site's design tokens (colors, fonts, spacing), TypeScript types for the content structures, and global assets (fonts, favicons). This is sequential and non-negotiable. Everything after this can be parallel.\n\n### 5. Extract How It Looks AND How It Behaves\n\nA website is not a screenshot — it's a living thing. Elements move, change, appear, and disappear in response to scrolling, hovering, clicking, resizing, and time. If you only extract the static CSS of each element, your clone will look right in a screenshot but feel dead when someone actually uses it.\n\nFor every element, extract its **appearance** (exact computed CSS via `getComputedStyle()`) AND its **behavior** (what changes, what triggers the change, and how the transition happens). Not \"it looks like 16px\" — extract the actual computed value. Not \"the nav changes on scroll\" — document the exact trigger (scroll position, IntersectionObserver threshold, viewport intersection), the before and after states (both sets of CSS values), and the transition (duration, easing, CSS transition vs. JS-driven vs. CSS `animation-timeline`).\n\nExamples of behaviors to watch for — these are illustrative, not exhaustive. The page may do things not on this list, and you must catch those too:\n- A navbar that shrinks, changes background, or gains a shadow after scrolling past a threshold\n- Elements that animate into view when they enter the viewport (fade-up, slide-in, stagger delays)\n- Sections that snap into place on scroll (`scroll-snap-type`)\n- Parallax layers that move at different rates than the scroll\n- Hover states that animate (not just change — the transition duration and easing matter)\n- Dropdowns, modals, accordions with enter/exit animations\n- Scroll-driven progress indicators or opacity transitions\n- Auto-playing carousels or cycling content\n- Dark-to-light (or any theme) transitions between page sections\n- **Tabbed/pill content that cycles** — buttons that switch visible card sets with transitions\n- **Scroll-driven tab/accordion switching** — sidebars where the active item auto-changes as content scrolls past (IntersectionObserver, NOT click handlers)\n- **Smooth scroll libraries** (Lenis, Locomotive Scroll) — check for `.lenis` class or scroll container wrappers\n\n### 6. Identify the Interaction Model Before Building\n\nThis is the single most expensive mistake in cloning: building a click-based UI when the original is scroll-driven, or vice versa. Before writing any builder prompt for an interactive section, you must definitively answer: **Is this section driven by clicks, scrolls, hovers, time, or some combination?**\n\nHow to determine this:\n1. **Don't click first.** Scroll through the section slowly and observe if things change on their own as you scroll.\n2. If they do, it's scroll-driven. Extract the mechanism: `IntersectionObserver`, `scroll-snap`, `position: sticky`, `animation-timeline`, or JS scroll listeners.\n3. If nothing changes on scroll, THEN click/hover to test for click/hover-driven interactivity.\n4. Document the interaction model explicitly in the component spec: \"INTERACTION MODEL: scroll-driven with IntersectionObserver\" or \"INTERACTION MODEL: click-to-switch with opacity transition.\"\n\nA section with a sticky sidebar and scrolling content panels is fundamentally different from a tabbed interface where clicking switches content. Getting this wrong means a complete rewrite, not a CSS tweak.\n\n### 7. Extract Every State, Not Just the Default\n\nMany components have multiple visual states — a tab bar shows different cards per tab, a header looks different at scroll position 0 vs 100, a card has hover effects. You must extract ALL states, not just whatever is visible on page load.\n\nFor tabbed/stateful content:\n- Click each tab/button via browser MCP\n- Extract the content, images, and card data for EACH state\n- Record which content belongs to which state\n- Note the transition animation between states (opacity, slide, fade, etc.)\n\nFor scroll-dependent elements:\n- Capture computed styles at scroll position 0 (initial state)\n- Scroll past the trigger threshold and capture computed styles again (scrolled state)\n- Diff the two to identify exactly which CSS properties change\n- Record the transition CSS (duration, easing, properties)\n- Record the exact trigger threshold (scroll position in px, or viewport intersection ratio)\n\n### 8. Spec Files Are the Source of Truth\n\nEvery component gets a specification file in `docs/research/components/` BEFORE any builder is dispatched. This file is the contract between your extraction work and the builder agent. The builder receives the spec file contents inline in its prompt — the file also persists as an auditable artifact that the user (or you) can review if something looks wrong.\n\nThe spec file is not optional. It is not a nice-to-have. If you dispatch a builder without first writing a spec file, you are shipping incomplete instructions based on whatever you can remember from a browser MCP session, and the builder will guess to fill gaps.\n\n### 9. Build Must Always Compile\n\nEvery builder agent must verify `npx tsc --noEmit` passes before finishing. After merging worktrees, you verify `npm run build` passes. A broken build is never acceptable, even temporarily.\n\n## Phase 1: Reconnaissance\n\nNavigate to the target URL with browser MCP.\n\nFollow **§0 (Visual crawl priority)** during the entire reconnaissance pass: images and backgrounds → SVGs/icons → motion/animations — before spending time on secondary copy tweaks.\n\n### Screenshots\n- Take **full-page screenshots** at desktop (1440px) and mobile (390px) viewports\n- Save to `docs/design-references/` with descriptive names\n- These are your master reference — builders will receive section-specific crops/screenshots later\n\n### Global Extraction\nExtract these from the page before doing anything else:\n\n**Fonts** — Inspect `<link>` tags for Google Fonts or self-hosted fonts. Check computed `font-family` on key elements (headings, body, code, labels). Document every family, weight, and style actually used. Configure them in `src/app/layout.tsx` using `next/font/google` or `next/font/local`.\n\n**Colors** — Extract the site's color palette from computed styles across the page. Update `src/app/globals.css` with the target's actual colors in the `:root` and `.dark` CSS variable blocks. Map them to shadcn's token names (background, foreground, primary, muted, etc.) where they fit. Add custom properties for colors that don't map to shadcn tokens.\n\n**Favicons & Meta** — Download favicons, apple-touch-icons, OG images, webmanifest to `public/seo/`. Update `layout.tsx` metadata.\n\n**Global UI patterns** — Identify any site-wide CSS or JS: custom scrollbar hiding, scroll-snap on the page container, global keyframe animations, backdrop filters, gradients used as overlays, **smooth scroll libraries** (Lenis, Locomotive Scroll — check for `.lenis`, `.locomotive-scroll`, or custom scroll container classes). Add these to `globals.css` and note any libraries that need to be installed.\n\n### Mandatory Interaction Sweep\n\nThis is a dedicated pass AFTER screenshots and BEFORE anything else. Its purpose is to discover every behavior on the page — many of which are invisible in a static screenshot.\n\n**Scroll sweep:** Scroll the page slowly from top to bottom via browser MCP. At each section, pause and observe:\n- Does the header change appearance? Record the scroll position where it triggers.\n- Do elements animate into view? Record which ones and the animation type.\n- Does a sidebar or tab indicator auto-switch as you scroll? Record the mechanism.\n- Are there scroll-snap points? Record which containers.\n- Is there a smooth scroll library active? Check for non-native scroll behavior.\n\n**Click sweep:** Click every element that looks interactive:\n- Every button, tab, pill, link, card\n- Record what happens: does content change? Does a modal open? Does a dropdown appear?\n- For tabs/pills: click EACH ONE and record the content that appears for each state\n\n**Hover sweep:** Hover over every element that might have hover states:\n- Buttons, cards, links, images, nav items\n- Record what changes: color, scale, shadow, underline, opacity\n\n**Responsive sweep:** Test at 3 viewport widths via browser MCP:\n- Desktop: 1440px\n- Tablet: 768px\n- Mobile: 390px\n- At each width, note which sections change layout (column → stack, sidebar disappears, etc.) and at approximately which breakpoint the change occurs.\n\nRe-check **all three widths** again in **Phase 5: Visual QA Diff** so tablet regressions are not skipped.\n\nSave all findings to `docs/research/BEHAVIORS.md`. This is your behavior bible — reference it when writing every component spec.\n\n### Chrome MCP: precise motion & animation extraction\n\nUse **evaluate script** (or equivalent) in Chrome MCP **during** the scroll/click/hover sweeps so you capture **live computed values**, not guesses from DevTools screenshots alone.\n\n**Rules:**\n\n- **No hand-waving** — if the reference uses `0.45s cubic-bezier(0.4, 0, 0.2, 1)`, the spec and implementation use that exact string (or Tailwind utilities **only** when the **resolved computed** animation/transition longhands match — class names alone are not proof).\n- **`@keyframes` source of truth** — for each animated element, record `getComputedStyle(el).animationName` and resolve the **full `@keyframes` rule** from stylesheets when allowed. Many sites inline CSS or use same-origin sheets: iterate `document.styleSheets` / `cssRules` inside `try/catch`; for **cross-origin** sheets that throw on `cssRules`, copy the rule from the **Network** response or **Sources** panel in Chrome and paste into `docs/research/MOTION.md` / the component spec. Never substitute a different easing or keyframe shape “because it looks close.”\n- **Scroll-driven animations** — for `animation-timeline: view()` / named timelines, capture **`view-timeline` / `scroll-timeline` on ancestors**, `animation-range`, `scroll-padding` / snap containers, and **which scroll container** is the timeline root (often `html`, sometimes a nested `overflow-y: auto` div). **`getComputedStyle` support varies** for scroll-animation longhands; also read **`cs.getPropertyValue('view-timeline-name')`**, **`scroll-timeline-name`**, **`animation-timeline`**, etc., when camelCase mirrors are missing in the engine you’re using.\n- **Before/after pairs** — for scroll- or hover-driven motion, capture computed `transform`, `opacity`, and longhand `animation-*` / `transition-*` **in both states** (see §7) and the **exact trigger** (px, ratio, or event).\n- **Libraries** — if Lenis/GSAP/Framer is present, document **version/hooks** from **script `src` URLs** (path often includes version), `//# sourceMappingURL` / comment banners in the bundle if visible, global identifiers on `window` when safe to read, **npm lockfile** only if this repo vendors that script, and **class/DOM hooks** (e.g. `.lenis`). Sample **computed style** at rest vs. mid-animation after interactions.\n- **Deliverable** — maintain `docs/research/MOTION.md`: page-level keyframes inventory, global scroll/smooth-scroll setup, and per-section pointers into component specs. Large pages may scope the audit script to a **section root selector** to avoid noise.\n\n**Motion audit script** (run via Chrome MCP; optional `rootSelector` limits to a subtree):\n\n```javascript\n(function motionAudit(rootSelector) {\n  const root = rootSelector ? document.querySelector(rootSelector) : document.body;\n  if (!root) return JSON.stringify({ error: 'root not found: ' + rootSelector });\n\n  function safeRules(sheet) {\n    try {\n      return [...sheet.cssRules];\n    } catch {\n      return [];\n    }\n  }\n\n  const walkKeyframes = (rules, out) => {\n    for (const rule of rules) {\n      if (rule.type === CSSRule.KEYFRAMES_RULE) {\n        out.push({ name: rule.name, cssText: rule.cssText });\n      } else if (rule.type === CSSRule.MEDIA_RULE && rule.cssRules) {\n        walkKeyframes(rule.cssRules, out);\n      }\n    }\n  };\n\n  const keyframes = [];\n  for (const sheet of document.styleSheets) {\n    walkKeyframes(safeRules(sheet), keyframes);\n  }\n\n  const animated = [];\n  // Include `root` — querySelectorAll('*') misses animations on the section container itself\n  const scope = [root, ...root.querySelectorAll('*')];\n  for (const el of scope) {\n    const cs = getComputedStyle(el);\n    const animName = cs.animationName;\n    const hasAnim = animName && animName !== 'none';\n    const transDurs = (cs.transitionDuration || '')\n      .split(',')\n      .map((s) => s.trim())\n      .filter(Boolean);\n    const hasTrans =\n      cs.transitionProperty &&\n      cs.transitionProperty !== 'none' &&\n      transDurs.some((d) => d !== '0s' && d !== '0ms');\n    if (!hasAnim && !hasTrans) continue;\n\n    animated.push({\n      hint: el.tagName.toLowerCase() + (el.className\n        ? '.' + String(el.className).trim().split(/\\s+/).slice(0, 4).join('.')\n        : ''),\n      animation: cs.animation,\n      animationName: cs.animationName,\n      animationDuration: cs.animationDuration,\n      animationTimingFunction: cs.animationTimingFunction,\n      animationDelay: cs.animationDelay,\n      animationIterationCount: cs.animationIterationCount,\n      animationDirection: cs.animationDirection,\n      animationFillMode: cs.animationFillMode,\n      animationPlayState: cs.animationPlayState,\n      animationTimeline: cs.animationTimeline,\n      animationRange: cs.animationRange,\n      transition: cs.transition,\n      transitionProperty: cs.transitionProperty,\n      transitionDuration: cs.transitionDuration,\n      transitionTimingFunction: cs.transitionTimingFunction,\n      transitionDelay: cs.transitionDelay,\n      transform: cs.transform,\n      transformOrigin: cs.transformOrigin,\n      opacity: cs.opacity,\n      willChange: cs.willChange,\n      viewTimelineName: cs.viewTimelineName,\n      scrollTimelineName: cs.scrollTimelineName,\n      viewTimelineNameRaw: cs.getPropertyValue('view-timeline-name'),\n      scrollTimelineNameRaw: cs.getPropertyValue('scroll-timeline-name')\n    });\n  }\n\n  return JSON.stringify(\n    {\n      prefersReducedMotion: window.matchMedia('(prefers-reduced-motion: reduce)').matches,\n      keyframes,\n      animatedElements: animated\n    },\n    null,\n    2\n  );\n})(''); // Pass section selector string, e.g. 'section.hero', instead of ''\n```\n\n### Page Topology\nMap out every distinct section of the page from top to bottom. Give each a working name. Document:\n- Their visual order\n- Which are fixed/sticky overlays vs. flow content\n- The overall page layout (scroll container, column structure, z-index layers)\n- Dependencies between sections (e.g., a floating nav that overlays everything)\n- **The interaction model** of each section (static, click-driven, scroll-driven, time-driven)\n\nSave this as `docs/research/PAGE_TOPOLOGY.md` — it becomes your assembly blueprint.\n\n## Phase 2: Foundation Build\n\nThis is sequential. Do it yourself (not delegated to an agent) since it touches many files:\n\n1. **Update fonts** in `layout.tsx` to match the target site's actual fonts\n2. **Update globals.css** with the target's color tokens, spacing values, keyframe animations, utility classes, and any **global scroll behaviors** (Lenis, smooth scroll CSS, scroll-snap on body)\n3. **Create TypeScript interfaces** in `src/types/` for the content structures you've observed\n4. **Extract SVG icons** — find all inline `<svg>` elements on the page, deduplicate them, and save as named React components in `src/components/icons.tsx`. Name them by visual function (e.g., `SearchIcon`, `ArrowRightIcon`, `LogoIcon`).\n5. **Download global assets** — write and run a Node.js script (`scripts/download-assets.mjs`) that downloads all images, videos, and other binary assets from the page to `public/`. Preserve meaningful directory structure.\n6. Verify: `npm run build` passes\n\n### Asset Discovery Script Pattern\n\nUse browser MCP to enumerate all assets on the page:\n\n```javascript\n// Run this via browser MCP to discover all assets\nJSON.stringify({\n  images: [...document.querySelectorAll('img')].map(img => ({\n    src: img.src || img.currentSrc,\n    alt: img.alt,\n    width: img.naturalWidth,\n    height: img.naturalHeight,\n    // Include parent info to detect layered compositions\n    parentClasses: img.parentElement?.className,\n    siblings: img.parentElement ? [...img.parentElement.querySelectorAll('img')].length : 0,\n    position: getComputedStyle(img).position,\n    zIndex: getComputedStyle(img).zIndex\n  })),\n  videos: [...document.querySelectorAll('video')].map(v => ({\n    src: v.src || v.querySelector('source')?.src,\n    poster: v.poster,\n    autoplay: v.autoplay,\n    loop: v.loop,\n    muted: v.muted\n  })),\n  backgroundImages: [...document.querySelectorAll('*')].filter(el => {\n    const bg = getComputedStyle(el).backgroundImage;\n    return bg && bg !== 'none';\n  }).map(el => ({\n    url: getComputedStyle(el).backgroundImage,\n    element: el.tagName + '.' + el.className?.split(' ')[0]\n  })),\n  svgCount: document.querySelectorAll('svg').length,\n  fonts: [...new Set([...document.querySelectorAll('*')].slice(0, 200).map(el => getComputedStyle(el).fontFamily))],\n  favicons: [...document.querySelectorAll('link[rel*=\"icon\"]')].map(l => ({ href: l.href, sizes: l.sizes?.toString() }))\n});\n```\n\nThen write a download script that fetches everything to `public/`. Use batched parallel downloads (4 at a time) with proper error handling.\n\n## Phase 3: Component Specification & Dispatch\n\nThis is the core loop. For each section in your page topology (top to bottom), you do THREE things: **extract**, **write the spec file**, then **dispatch builders**.\n\n### Step 1: Extract\n\nFor each section, use browser MCP to extract everything:\n\n1. **Screenshot** the section in isolation (scroll to it, screenshot the viewport). Save to `docs/design-references/`.\n\n2. **Extract CSS** for every element in the section. Use the extraction script below — don't hand-measure individual properties. Run it **once per spec scope**: usually the **section wrapper** selector (one subtree). For very large sections, run again on **inner selectors** (e.g. each card type) — the `depth` / `children.slice(0, 20)` limits are **deliberately capped** for MCP payload size; raise them or narrow `SELECTOR` when the spec is incomplete.\n\n```javascript\n// Per-component extraction — run via browser MCP\n// Replace SELECTOR with the actual CSS selector for the component\n(function(selector) {\n  const el = document.querySelector(selector);\n  if (!el) return JSON.stringify({ error: 'Element not found: ' + selector });\n  const props = [\n    'fontSize','fontWeight','fontFamily','lineHeight','letterSpacing','color',\n    'textTransform','textDecoration','backgroundColor','background',\n    'padding','paddingTop','paddingRight','paddingBottom','paddingLeft',\n    'margin','marginTop','marginRight','marginBottom','marginLeft',\n    'width','height','maxWidth','minWidth','maxHeight','minHeight',\n    'display','flexDirection','justifyContent','alignItems','gap',\n    'gridTemplateColumns','gridTemplateRows',\n    'borderRadius','border','borderTop','borderBottom','borderLeft','borderRight',\n    'boxShadow','overflow','overflowX','overflowY',\n    'position','top','right','bottom','left','zIndex',\n    'opacity','transform','transformOrigin','transition','transitionProperty','transitionDuration','transitionTimingFunction','transitionDelay',\n    'animation','animationName','animationDuration','animationTimingFunction','animationDelay','animationIterationCount','animationDirection','animationFillMode','animationPlayState','animationTimeline','animationRange','viewTimelineName','scrollTimelineName','cursor',\n    'objectFit','objectPosition','mixBlendMode','filter','backdropFilter',\n    'whiteSpace','textOverflow','WebkitLineClamp'\n  ];\n  function extractStyles(element) {\n    const cs = getComputedStyle(element);\n    const styles = {};\n    // Heuristic filter — for layout-audits, re-add any dropped `auto` / `normal` / transparent values the spec calls out as significant\n    props.forEach(p => { const v = cs[p]; if (v && v !== 'none' && v !== 'normal' && v !== 'auto' && v !== '0px' && v !== 'rgba(0, 0, 0, 0)') styles[p] = v; });\n    return styles;\n  }\n  function walk(element, depth) {\n    if (depth > 4) return null;\n    const children = [...element.children];\n    return {\n      tag: element.tagName.toLowerCase(),\n      classes: element.className?.toString().split(' ').slice(0, 5).join(' '),\n      text: element.childNodes.length === 1 && element.childNodes[0].nodeType === 3 ? element.textContent.trim().slice(0, 200) : null,\n      styles: extractStyles(element),\n      images: element.tagName === 'IMG' ? { src: element.src, alt: element.alt, naturalWidth: element.naturalWidth, naturalHeight: element.naturalHeight } : null,\n      childCount: children.length,\n      children: children.slice(0, 20).map(c => walk(c, depth + 1)).filter(Boolean)\n    };\n  }\n  return JSON.stringify(walk(el, 0), null, 2);\n})('SELECTOR');\n```\n\n3. **Extract multi-state styles** — for any element with multiple states (scroll-triggered, hover, active tab), capture BOTH states:\n\n```javascript\n// State A: capture styles at current state (e.g., scroll position 0)\n// Then trigger the state change (scroll, click, hover via browser MCP)\n// State B: re-run the extraction script on the same element\n// The diff between A and B IS the behavior specification\n```\n\nRecord the diff explicitly: \"Property X changes from VALUE_A to VALUE_B, triggered by TRIGGER, with transition: TRANSITION_CSS.\"\n\n4. **Extract real content** — all text, alt attributes, aria labels, placeholder text. Use `element.textContent` for each text node. For tabbed/stateful content, **click each tab and extract content per state**.\n\n5. **Identify assets** this section uses — which downloaded images/videos from `public/`, which icon components from `icons.tsx`. Check for **layered images** (multiple `<img>` or background-images stacked in the same container).\n\n6. **Assess complexity** — how many distinct sub-components does this section contain? A distinct sub-component is an element with its own unique styling, structure, and behavior (e.g., a card, a nav item, a search panel).\n\n### Step 2: Write the Component Spec File\n\nFor each section (or sub-component, if you're breaking it up), create a spec file in `docs/research/components/`. This is NOT optional — every builder must have a corresponding spec file.\n\n**File path:** `docs/research/components/<component-name>.spec.md`\n\n**Template:**\n\n```markdown\n# <ComponentName> Specification\n\n## Overview\n- **Target file:** `src/components/<ComponentName>.tsx`\n- **Screenshot:** `docs/design-references/<screenshot-name>.png`\n- **Interaction model:** <static | click-driven | scroll-driven | time-driven>\n\n## DOM Structure\n<Describe the element hierarchy — what contains what>\n\n## Computed Styles (exact values from getComputedStyle)\n\n### Container\n- display: ...\n- padding: ...\n- maxWidth: ...\n- (every relevant property with exact values)\n\n### <Child element 1>\n- fontSize: ...\n- color: ...\n- (every relevant property)\n\n### <Child element N>\n...\n\n## Motion & animation (exact — from Chrome MCP `getComputedStyle` + stylesheet rules)\n\n- **Keyframe names in use:** ...\n- **Full `@keyframes` blocks** (paste `cssText` or equivalent); note **cross-origin** gaps and how you filled them\n- **Per-element:** shorthand/longhand `animation` and `transition` values **verbatim** for resting state\n- **Scroll/view timelines:** timeline name, range, scroll container, snap/observer thresholds\n- **Libraries:** Lenis / GSAP / Framer / Lottie — hooks, init pattern, timing copied from reference\n- **`prefers-reduced-motion`:** matched behavior on the reference\n\n## States & Behaviors\n\n### <Behavior name, e.g., \"Scroll-triggered floating mode\">\n- **Trigger:** <exact mechanism — scroll position 50px, IntersectionObserver rootMargin \"-30% 0px\", click on .tab-button, hover>\n- **State A (before):** maxWidth: 100vw, boxShadow: none, borderRadius: 0\n- **State B (after):** maxWidth: 1200px, boxShadow: 0 4px 20px rgba(0,0,0,0.1), borderRadius: 16px\n- **Transition:** transition: all 0.3s ease\n- **Implementation approach:** <CSS transition + scroll listener | IntersectionObserver | CSS animation-timeline | etc.>\n\n### Hover states\n- **<Element>:** <property>: <before> → <after>, transition: <value>\n\n## Per-State Content (if applicable)\n\n### State: \"Featured\"\n- Title: \"...\"\n- Subtitle: \"...\"\n- Cards: [{ title, description, image, link }, ...]\n\n### State: \"Productivity\"\n- Title: \"...\"\n- Cards: [...]\n\n## Assets\n- Background image: `public/images/<file>.webp`\n- Overlay image: `public/images/<file>.png`\n- Icons used: <ArrowIcon>, <SearchIcon> from icons.tsx\n\n## Text Content (verbatim)\n<All text content, copy-pasted from the live site>\n\n## Responsive Behavior\n- **Desktop (1440px):** <layout description>\n- **Tablet (768px):** <what changes — e.g., \"maintains 2-column, gap reduces to 16px\">\n- **Mobile (390px):** <what changes — e.g., \"stacks to single column, images full-width\">\n- **Breakpoint:** layout switches at ~<N>px\n```\n\nFill every section. If a section doesn't apply (e.g., no states for a static footer), write \"N/A\" — but think twice before marking States & Behaviors as N/A. Even a footer might have hover states on links.\n\n### Step 3: Dispatch Builders\n\nBased on complexity, dispatch builder agent(s) in worktree(s):\n\n**Simple section** (1-2 sub-components): One builder agent gets the entire section.\n\n**Complex section** (3+ distinct sub-components): Break it up. One agent per sub-component, plus one agent for the section wrapper that imports them. Sub-component builders go first since the wrapper depends on them.\n\n**What every builder agent receives:**\n- The full contents of its component spec file (inline in the prompt — don't say \"go read the spec file\")\n- Path to the section screenshot in `docs/design-references/`\n- Which shared components to import (`icons.tsx`, `cn()`, shadcn primitives)\n- The target file path (e.g., `src/components/HeroSection.tsx`)\n- Instruction to verify with `npx tsc --noEmit` before finishing\n- For responsive behavior: the specific breakpoint values and what changes\n\n**Don't wait.** As soon as you've dispatched the builder(s) for one section, move to extracting the next section. Builders work in parallel in their worktrees while you continue extraction.\n\n### Step 4: Merge\n\nAs builder agents complete their work:\n- Merge their worktree branches into main\n- You have full context on what each agent built, so resolve any conflicts intelligently\n- After each merge, verify the build still passes: `npm run build`\n- If a merge introduces type errors, fix them immediately\n\nThe extract → spec → dispatch → merge cycle continues until all sections are built.\n\n## Phase 4: Page Assembly\n\nAfter all sections are built and merged, wire everything together in `src/app/page.tsx`:\n\n- Import all section components\n- Implement the page-level layout from your topology doc (scroll containers, column structures, sticky positioning, z-index layering)\n- Connect real content to component props\n- Implement page-level behaviors: scroll snap, scroll-driven animations, dark-to-light transitions, intersection observers, smooth scroll (Lenis etc.)\n- Verify: `npm run build` passes clean\n\n## Phase 5: Visual QA Diff\n\nAfter assembly, do NOT declare the clone complete. Take side-by-side comparison screenshots:\n\n1. Open the original site and your clone side-by-side (or take screenshots at the same viewport widths)\n2. Compare section by section, top to bottom, at desktop (1440px)\n3. Compare again at tablet (768px) — same widths as **Responsive sweep**\n4. Compare again at mobile (390px)\n5. For each discrepancy found:\n   - Check the component spec file — was the value extracted correctly?\n   - If the spec was wrong: re-extract from browser MCP, update the spec, fix the component\n   - If the spec was right but the builder got it wrong: fix the component to match the spec\n6. Test all interactive behaviors: scroll through the page, click every button/tab, hover over interactive elements\n7. **Motion QA (Chrome MCP again if needed):** For every animated or transitioned element, re-run `getComputedStyle` on the reference vs. the clone at the **same scroll position / interaction state** and compare **`animation-*`, `transition-*`, `transform`, `opacity`** longhands — **resolved** durations/easings must match where the cascade specifies them (Tailwind classes are fine when they compile to the same longhands; verify in **Computed**, not by class name alone). Re-record any `@keyframes` the reference still exposes so the clone can be patched until timings align.\n8. Verify smooth scroll feels right, header transitions work, tab switching works, stagger and scroll-driven reveals match the reference\n\nOnly after this visual + motion QA pass is the clone complete.\n\n## Pre-Dispatch Checklist\n\nBefore dispatching ANY builder agent, verify you can check every box. If you can't, go back and extract more.\n\n- [ ] Spec file written to `docs/research/components/<name>.spec.md` with ALL sections filled\n- [ ] Every CSS value in the spec is from `getComputedStyle()`, not estimated\n- [ ] Interaction model is identified and documented (static / click / scroll / time)\n- [ ] For stateful components: every state's content and styles are captured\n- [ ] For scroll-driven components: trigger threshold, before/after styles, and transition are recorded\n- [ ] For hover states: before/after values and transition timing are recorded\n- [ ] Motion: relevant `@keyframes` captured (or cross-origin gap documented with manual paste), `docs/research/MOTION.md` updated, component specs include **exact** `animation*` / `transition*` / timeline fields from Chrome MCP\n- [ ] All images in the section are identified (including overlays and layered compositions)\n- [ ] Responsive behavior is documented for desktop, tablet, and mobile (1440 / 768 / 390)\n- [ ] Text: **verbatim** from the reference **except** hero, main headings, and primary CTAs — those match the **SaaS idea** (reference copy still captured for audit if useful)\n- [ ] The builder prompt is under ~150 lines of spec; if over, the section needs to be split\n\n## What NOT to Do\n\nThese are lessons from previous failed clones — each one cost hours of rework:\n\n- **Don't build click-based tabs when the original is scroll-driven (or vice versa).** Determine the interaction model FIRST by scrolling before clicking. This is the #1 most expensive mistake — it requires a complete rewrite, not a CSS fix.\n- **Don't extract only the default state.** If there are tabs showing \"Featured\" on load, click Productivity, Creative, Lifestyle and extract each one's cards/content. If the header changes on scroll, capture styles at position 0 AND position 100+.\n- **Don't miss overlay/layered images.** A background watercolor + foreground UI mockup = 2 images. Check every container's DOM tree for multiple `<img>` elements and positioned overlays.\n- **Don't build mockup components for content that's actually videos/animations.** Check if a section uses `<video>`, Lottie, or canvas before building elaborate HTML mockups of what the video shows.\n- **Don't approximate CSS classes.** \"It looks like `text-lg`\" is wrong if the computed value is `18px` and `text-lg` is `18px/28px` but the actual line-height is `24px`. Extract exact values.\n- **Don't build everything in one monolithic commit.** The whole point of this pipeline is incremental progress with verified builds at each step.\n- **Don't reference docs from builder prompts.** Each builder gets the CSS spec inline in its prompt — never \"see DESIGN_TOKENS.md for colors.\" The builder should have zero need to read external docs.\n- **Don't skip asset extraction.** Without real images, videos, and fonts, the clone will always look fake regardless of how perfect the CSS is.\n- **Don't give a builder agent too much scope.** If you're writing a builder prompt and it's getting long because the section is complex, that's a signal to break it into smaller tasks.\n- **Don't bundle unrelated sections into one agent.** A CTA section and a footer are different components with different designs — don't hand them both to one agent and hope for the best.\n- **Don't skip responsive extraction.** If you only inspect at desktop width, the clone will break at tablet and mobile. Test at 1440, 768, and 390 during extraction.\n- **Don't forget smooth scroll libraries.** Check for Lenis (`.lenis` class), Locomotive Scroll, or similar. Default browser scrolling feels noticeably different and the user will spot it immediately.\n- **Don't dispatch builders without a spec file.** The spec file forces exhaustive extraction and creates an auditable artifact. Skipping it means the builder gets whatever you can fit in a prompt from memory.\n\n## Completion\n\nWhen done, report:\n- Total sections built\n- Total components created\n- Total spec files written (should match components)\n- Total assets downloaded (images, videos, SVGs, fonts)\n- Build status (`npm run build` result)\n- Visual QA results (any remaining discrepancies)\n- Any known gaps or limitations\n",
+  "prompt": "\n# Launchframe\n\n**`/launchframe`** is the **only** slash command for reverse-engineering a live site into this codebase. Parse **the reference URL(s) and SaaS idea the user passed with /launchframe**:\n\n1. **Reference URL(s)** — Every `http://` or `https://` token (clone each site; use `docs/research/<hostname>/` when there are multiple).\n2. **SaaS idea** — All remaining text that is not a URL (often quoted): product pitch for hero and key marketing lines.\n\nYou are a **foreman walking the job site** — write specs per section, dispatch builders in parallel where possible, merge, QA. Not a shallow inspect-then-guess flow.\n\n## Step 0 — Persist Launchframe inputs\n\nBefore reconnaissance, write or update:\n\n- `src/lib/launchframe-config.ts` — `LAUNCHFRAME_SOURCE_URL` (primary URL, usually first), `LAUNCHFRAME_SAAS_IDEA`\n- `launchframe.context.json`\n- `docs/research/LAUNCHFRAME.md` — URLs and SaaS idea\n\n## SaaS copy overlay (Phase 4 assembly and final polish)\n\nAfter structure and styles match the reference, apply the **SaaS idea** to hero, headings, and primary CTAs where the reference uses interchangeable marketing copy, **without** changing layout grids, spacing, or motion from extracted specs. **Brand identity** (below) must be **original** for anything you ship as the user’s product — never pass off the reference company’s trademarks or distinctive marks.\n\n---\n\n## Scope Defaults\n\nThe target page(s) are the URL(s) you parsed in Pre-Flight. Clone exactly what's visible at each URL. Unless the user specifies otherwise, use these defaults:\n\n- **Fidelity level:** Pixel-perfect — exact match in colors, spacing, typography, animations\n- **In scope:** Visual layout and styling, component structure and interactions, responsive design, mock data for demo purposes\n- **Out of scope:** Real backend / database, authentication, real-time features, SEO optimization, accessibility audit\n- **Customization:** Structure and visuals — pure emulation of the reference. **Marketing copy** — apply the parsed **SaaS idea** where hero/headlines/CTAs are interchangeable (see “SaaS copy overlay” above); do not invent a different product than the user’s idea.\n- **Brand identity:** The reference is a **pattern** for layout, motion, and UI craft — **not** permission to ship their brand. Unless the user **explicitly** asks for a faithful copy of the reference brand (e.g. licensed work, clearly labeled internal mock, private design audit), **invent an original brand** aligned with the SaaS idea: product name, wordmark or simple logomark (SVG or styled text) sized to the same logo slot, favicon / app icon, OG imagery, and a cohesive palette. Do **not** reuse their trademarked logo paths, mascot art, or distinctive illustrative brand assets; use originals or functional UI icons instead. Hero or lifestyle images that center the reference brand should be replaced with **original** imagery or neutral compositions that keep the **same layout rhythm**. Note in `docs/research/LAUNCHFRAME.md` which marks and assets are **original brand** versus **layout-only** extraction.\n\nIf the user provides additional instructions (specific fidelity level, customizations, extra context), honor those over the defaults.\n\n## Pre-Flight\n\n1. **Browser automation is required.** Check for available browser MCP tools (Chrome MCP, Playwright MCP, Browserbase MCP, Puppeteer MCP, etc.). Use whichever is available — if multiple exist, **prefer Chrome DevTools MCP** for inspection: it maps to the same engine the user sees and preserves **computed `animation` / `transition` / scroll-driven** values accurately. If none are detected, ask the user which browser tool they have and how to connect it. This skill cannot work without browser automation. **Pixel-perfect MCP:** every extract and QA pass must follow **Browser MCP: pixel-perfect capture** (Phase 1) — locked viewports, **verbatim** `getComputedStyle` strings (including fractional `px`, `rem`, `oklch`/`rgba`), and **geometry** from `getBoundingClientRect()` where layout matters. **Motion parity:** every pass must treat **motion as first-class** — same **durations, delays, easings (`cubic-bezier` / steps), iteration counts, fill modes, directions, keyframe percentages, and scroll/view-timeline bindings** as the live site, not “similar” motion.\n2. **Parse arguments** — extract every `http://` / `https://` URL token (there may be several). **SaaS idea** = the remaining non-URL text (trim outer quotes). Normalize and validate each URL; if any are invalid, or the SaaS idea is missing, ask the user once. For each valid URL, verify it is accessible via your browser MCP tool.\n3. Verify the base project builds: `npm run build`. The Next.js + shadcn/ui + Tailwind v4 scaffold should already be in place. If not, tell the user to set it up first.\n4. Create the output directories if they don't exist: `docs/research/`, `docs/research/components/`, `docs/design-references/`, `scripts/`. For multiple clones, also prepare per-site folders like `docs/research/<hostname>/` and `docs/design-references/<hostname>/`.\n5. When working with multiple sites in one command, optionally confirm whether to run them in parallel (recommended, if resources allow) or sequentially to avoid overload.\n\n## Guiding Principles\n\nThese are the truths that separate a successful clone from a \"close enough\" mess. Internalize them — they should inform every decision you make.\n\n### 0. Visual crawl priority (images, SVGs, motion — first)\n\nWhen you traverse the DOM and the Network panel, do **not** treat all nodes equally. Work in this **priority order** so the clone feels like the original, not a wireframe:\n\n1. **Images (raster + video stills)** — Enumerate `<img>`, `<picture>`, responsive `srcset`, `data-*` lazy URLs, **computed `background-image`** on the element and parents (including pseudo-elements), mask images, `<video poster>`, hero media. **Scrape:** download binary assets to `public/images/` (or `public/videos/`) with stable paths referenced in specs. **Create** a replacement PNG/WebP/SVG only when the asset is blocked (CORS, auth cookie, 403), ephemeral, or impossible to URL-fetch — use a high-DPI screenshot crop, traced artwork, or CSS gradient approximation, and mark `ASSET_SOURCE: generated` in the component spec with a short reason.\n2. **SVGs & iconography** — Inline `<svg>`, sprite `symbol` defs, **SVG used as masks/filters**, icon fonts (prefer path extraction). Convert to `@/components/icons.tsx` (or section-local components) with meaningful names. Prioritize crisp edges and correct `viewBox` over shrinking bundle size during emulation.\n3. **Motion & animation** — CSS `@keyframes`, `animation`, `animation-timeline` / `view-timeline` / `animation-range`, `transition`, `transform`, `transform-origin`, will-change; JS-driven motion (carousel timing, IntersectionObserver reveals); libraries (GSAP, Framer, Lottie JSON, Lenis). Capture **numbers** (ms, `cubic-bezier()`, `steps()`, stagger, scroll thresholds, **named keyframe blocks**), not adjectives — paste **verbatim** `cssText` for relevant `@keyframes` into specs when obtainable. Include **reduced-motion** (`matchMedia('(prefers-reduced-motion: reduce)'))` behavior if present.\n\nOnly after the above are accounted for should you spend cycle time on minor text or non-visual refactors. A perfect grid with missing hero art and dead animation still fails the clone.\n\n### 1. Completeness Beats Speed\n\nEvery builder agent must receive **everything** it needs to do its job perfectly: screenshot, exact CSS values, downloaded assets with local paths, real text content, component structure. If a builder has to guess anything — a color, a font size, a padding value — you have failed at extraction. Take the extra minute to extract one more property rather than shipping an incomplete brief.\n\n### 2. Small Tasks, Perfect Results\n\nWhen an agent gets \"build the entire features section,\" it glosses over details — it approximates spacing, guesses font sizes, and produces something \"close enough\" but clearly wrong. When it gets a single focused component with exact CSS values, it nails it every time.\n\nLook at each section and judge its complexity. A simple banner with a heading and a button? One agent. A complex section with 3 different card variants, each with unique hover states and internal layouts? One agent per card variant plus one for the section wrapper. When in doubt, make it smaller.\n\n**Complexity budget rule:** If a builder prompt exceeds ~150 lines of spec content, the section is too complex for one agent. Break it into smaller pieces. This is a mechanical check — don't override it with \"but it's all related.\"\n\n### 3. Real Content, Real Assets\n\nExtract the actual text, images, videos, and SVGs from the live site. This is a clone, not a mockup. Use `element.textContent`, download every `<img>` and `<video>`, extract inline `<svg>` elements as React components. The only time you generate content is when something is clearly server-generated and unique per session.\n\n**Prioritize** (see §0): downloadable imagery and backgrounds first, then SVG/icon layers, then motion. If you must **fabricate** an asset, prefer screenshot-based exports or traced vectors tied to measured box sizes — avoid unrelated stock art.\n\n**Layered assets matter.** A section that looks like one image is often multiple layers — a background watercolor/gradient, a foreground UI mockup PNG, an overlay icon. Inspect each container's full DOM tree and enumerate ALL `<img>` elements and background images within it, including absolutely-positioned overlays. Missing an overlay image makes the clone look empty even if the background is correct.\n\n### 4. Foundation First\n\nNothing can be built until the foundation exists: global CSS with the target site's design tokens (colors, fonts, spacing), TypeScript types for the content structures, and global assets (fonts, favicons). This is sequential and non-negotiable. Everything after this can be parallel.\n\n### 5. Extract How It Looks AND How It Behaves\n\nA website is not a screenshot — it's a living thing. Elements move, change, appear, and disappear in response to scrolling, hovering, clicking, resizing, and time. If you only extract the static CSS of each element, your clone will look right in a screenshot but feel dead when someone actually uses it.\n\nFor every element, extract its **appearance** (exact computed CSS via `getComputedStyle()`) AND its **behavior** (what changes, what triggers the change, and how the transition happens). Not \"it looks like 16px\" — extract the actual computed value. Not \"the nav changes on scroll\" — document the exact trigger (scroll position, IntersectionObserver threshold, viewport intersection), the before and after states (both sets of CSS values), and the transition (duration, easing, CSS transition vs. JS-driven vs. CSS `animation-timeline`).\n\nExamples of behaviors to watch for — these are illustrative, not exhaustive. The page may do things not on this list, and you must catch those too:\n- A navbar that shrinks, changes background, or gains a shadow after scrolling past a threshold\n- Elements that animate into view when they enter the viewport (fade-up, slide-in, stagger delays)\n- Sections that snap into place on scroll (`scroll-snap-type`)\n- Parallax layers that move at different rates than the scroll\n- Hover states that animate (not just change — the transition duration and easing matter)\n- Dropdowns, modals, accordions with enter/exit animations\n- Scroll-driven progress indicators or opacity transitions\n- Auto-playing carousels or cycling content\n- Dark-to-light (or any theme) transitions between page sections\n- **Tabbed/pill content that cycles** — buttons that switch visible card sets with transitions\n- **Scroll-driven tab/accordion switching** — sidebars where the active item auto-changes as content scrolls past (IntersectionObserver, NOT click handlers)\n- **Smooth scroll libraries** (Lenis, Locomotive Scroll) — check for `.lenis` class or scroll container wrappers\n\n### 6. Identify the Interaction Model Before Building\n\nThis is the single most expensive mistake in cloning: building a click-based UI when the original is scroll-driven, or vice versa. Before writing any builder prompt for an interactive section, you must definitively answer: **Is this section driven by clicks, scrolls, hovers, time, or some combination?**\n\nHow to determine this:\n1. **Don't click first.** Scroll through the section slowly and observe if things change on their own as you scroll.\n2. If they do, it's scroll-driven. Extract the mechanism: `IntersectionObserver`, `scroll-snap`, `position: sticky`, `animation-timeline`, or JS scroll listeners.\n3. If nothing changes on scroll, THEN click/hover to test for click/hover-driven interactivity.\n4. Document the interaction model explicitly in the component spec: \"INTERACTION MODEL: scroll-driven with IntersectionObserver\" or \"INTERACTION MODEL: click-to-switch with opacity transition.\"\n\nA section with a sticky sidebar and scrolling content panels is fundamentally different from a tabbed interface where clicking switches content. Getting this wrong means a complete rewrite, not a CSS tweak.\n\n### 7. Extract Every State, Not Just the Default\n\nMany components have multiple visual states — a tab bar shows different cards per tab, a header looks different at scroll position 0 vs 100, a card has hover effects. You must extract ALL states, not just whatever is visible on page load.\n\nFor tabbed/stateful content:\n- Click each tab/button via browser MCP\n- Extract the content, images, and card data for EACH state\n- Record which content belongs to which state\n- Note the transition animation between states (opacity, slide, fade, etc.)\n\nFor scroll-dependent elements:\n- Capture computed styles at scroll position 0 (initial state)\n- Scroll past the trigger threshold and capture computed styles again (scrolled state)\n- Diff the two to identify exactly which CSS properties change\n- Record the transition CSS (duration, easing, properties)\n- Record the exact trigger threshold (scroll position in px, or viewport intersection ratio)\n\n### 8. Spec Files Are the Source of Truth\n\nEvery component gets a specification file in `docs/research/components/` BEFORE any builder is dispatched. This file is the contract between your extraction work and the builder agent. The builder receives the spec file contents inline in its prompt — the file also persists as an auditable artifact that the user (or you) can review if something looks wrong.\n\nThe spec file is not optional. It is not a nice-to-have. If you dispatch a builder without first writing a spec file, you are shipping incomplete instructions based on whatever you can remember from a browser MCP session, and the builder will guess to fill gaps.\n\n### 9. Build Must Always Compile\n\nEvery builder agent must verify `npx tsc --noEmit` passes before finishing. After merging worktrees, you verify `npm run build` passes. A broken build is never acceptable, even temporarily.\n\n## Phase 1: Reconnaissance\n\nNavigate to the target URL with browser MCP.\n\nFollow **§0 (Visual crawl priority)** during the entire reconnaissance pass: images and backgrounds → SVGs/icons → motion/animations — before spending time on secondary copy tweaks.\n\n### Multi-pass DOM reconnaissance (mandatory)\n\n**A single scroll is not enough.** You must run the live page **multiple full cycles** in browser MCP — each time walking the **DOM**, the **asset graph**, and the **motion flow** — until what you write in specs is **exactly** what the reference does, not a remembered approximation.\n\n**Minimum passes** (run **Pass 1–4 at 1440px first**; **repeat Pass 1–3 at 768px and 390px** wherever layout or node order differs; reconciliation can be width-specific):\n\n1. **Structure pass (DOM, top → bottom):** Traverse from **`document.body`** in **depth-first order** (or section-by-section using regions you will later name in `PAGE_TOPOLOGY.md`). For every node that affects layout or visible content, record **tag**, **selector hint** (stable `id` / first meaningful classes), **`childNodes` order** (siblings matter — flex/grid gaps and wrappers are not optional), landmarks, and **which element is the scroll container** (`overflow` / `overscroll-behavior`). Produce a **literal hierarchy** in `docs/research/DOM_SKETCH.md` (whole page or per-section files under `docs/research/`) so builders cannot silently reorder or collapse wrapper divs. **Include “invisible” structure** (spacers, gradient overlays, absolutely positioned siblings).\n2. **Asset pass (images & layered media):** **Second full traversal** focused on §0 priority 1–2: every `<img>`, `<picture>` / `srcset`, `<video>` / `poster`, computed **`background-image`** on the element **and** ancestors, CSS masks, inline SVG / sprites. Cross-check with the Network panel. Every URL must end up **downloaded to `public/`** with a documented path, or marked **blocked / generated** with a substitute and reason.\n3. **Motion & flow pass:** **Third full journey** — slow scroll **top to bottom**, then **scroll again** (different speed) to catch lazy-mounted nodes, staggered `IntersectionObserver` reveals, and scroll-snapped stops. Run **`motionAudit`** (below) **per major section root** after scrolling that section into view. **Re-do click/hover sweeps** here while sampling `getComputedStyle` so **transitions and keyframe-driven flow** match the reference’s **order and timing**, not a generic fade-in.\n4. **Reconciliation pass:** **Fourth pass** — read `DOM_SKETCH.md`, `BEHAVIORS.md`, asset inventory, and `MOTION.md` together; list **gaps** (missing wrapper, second image layer, unnamed animation, wrong child order). For every gap, return to the site and **repeat Pass 1–3** for that subtree until the spec is closed.\n\n**Hard rule:** Every fact in a component spec must trace to **which pass** produced it. If you “kind of remember” a structure from an earlier visit, **run the page again** — do not guess.\n\n### Screenshots\n- Take **full-page screenshots** at desktop (1440px) and mobile (390px) viewports\n- Save to `docs/design-references/` with descriptive names\n- These are your master reference — builders will receive section-specific crops/screenshots later\n\n### Browser MCP: pixel-perfect capture\n\nMCP-driven inspection must be **pixel-perfect**, not “close.” Apply this on **every** reconnaissance width and again in Phase 5.\n\n- **Viewport lock:** Resize the MCP-controlled window to exact **CSS widths** used for this project: **1440**, **768**, and **390** px (height large enough to avoid accidental mobile chrome quirks). After each resize, record **`window.innerWidth`**, **`document.documentElement.clientWidth`**, and **`devicePixelRatio`** (run the snippet below). Extraction and QA **use the same widths** so diffs are valid.\n- **Zoom:** Ensure **browser zoom is 100%** (not 90%/110%). Zoom breaks `px` parity and screenshots.\n- **Computed styles are verbatim:** When copying from `getComputedStyle()`, paste **exact returned strings** — including **fractional** lengths (`12.8px`), **negative** values, **`oklch()`/`lch()`/`rgba()`**, `letter-spacing`, `line-height`, `border`/`outline`, and **`calc()`**/`min()` that the cascade resolves to. **Do not round** for “clean” tokens (no `~16px`; no swapping `0.875rem` for `14px` unless the computed value truly equals that).\n- **Geometry:** When alignment, gutters, or sticky offsets matter, record **`getBoundingClientRect()`** (`x`, `y`, `width`, `height`) at the relevant scroll position and width — not only padding/margin strings. Fixed/sticky layers need **viewport-relative** positions.\n- **Pseudos & layers:** Extract **`::before` / `::after`** (and other pseudos the reference uses) with `getComputedStyle(el, '::before')` etc.; they often hold gradients, borders, or icons that define the pixel look.\n- **Anti-aliasing & text:** Prefer **real Chrome** via MCP so **font rasterization** matches; load the **same font files/weights** in Next.js as on the reference. Note `font-feature-settings` / `font-variation-settings` / `-webkit-font-smoothing` if non-default.\n- **Screenshots vs. DOM:** Screenshots are for **diffing**; the **spec values** still come from **computed style + rect**, not from eyeballing pixels.\n\n**Viewport diagnostic** (run via browser MCP after each resize):\n\n```javascript\nJSON.stringify({\n  innerWidth: window.innerWidth,\n  clientWidth: document.documentElement.clientWidth,\n  clientHeight: document.documentElement.clientHeight,\n  devicePixelRatio: window.devicePixelRatio,\n  scrollY: window.scrollY\n});\n```\n\n### Global Extraction\nExtract these from the page before doing anything else:\n\n**Fonts** — Inspect `<link>` tags for Google Fonts or self-hosted fonts. Check computed `font-family` on key elements (headings, body, code, labels). Document every family, weight, and style actually used. Configure them in `src/app/layout.tsx` using `next/font/google` or `next/font/local`.\n\n**Colors** — Extract the site's color palette from computed styles across the page. Update `src/app/globals.css` with the target's actual colors in the `:root` and `.dark` CSS variable blocks. Map them to shadcn's token names (background, foreground, primary, muted, etc.) where they fit. Add custom properties for colors that don't map to shadcn tokens.\n\n**Favicons & Meta** — Download favicons, apple-touch-icons, OG images, webmanifest to `public/seo/`. Update `layout.tsx` metadata.\n\n**Global UI patterns** — Identify any site-wide CSS or JS: custom scrollbar hiding, scroll-snap on the page container, global keyframe animations, backdrop filters, gradients used as overlays, **smooth scroll libraries** (Lenis, Locomotive Scroll — check for `.lenis`, `.locomotive-scroll`, or custom scroll container classes). Add these to `globals.css` and note any libraries that need to be installed.\n\n### Mandatory Interaction Sweep\n\nThis is a dedicated pass AFTER screenshots and BEFORE anything else. Its purpose is to discover every behavior on the page — many of which are invisible in a static screenshot.\n\n**Scroll sweep:** Scroll the page slowly from top to bottom via browser MCP. At each section, pause and observe:\n- Does the header change appearance? Record the scroll position where it triggers.\n- Do elements animate into view? Record which ones and the animation type.\n- Does a sidebar or tab indicator auto-switch as you scroll? Record the mechanism.\n- Are there scroll-snap points? Record which containers.\n- Is there a smooth scroll library active? Check for non-native scroll behavior.\n\n**Click sweep:** Click every element that looks interactive:\n- Every button, tab, pill, link, card\n- Record what happens: does content change? Does a modal open? Does a dropdown appear?\n- For tabs/pills: click EACH ONE and record the content that appears for each state\n\n**Hover sweep:** Hover over every element that might have hover states:\n- Buttons, cards, links, images, nav items\n- Record what changes: color, scale, shadow, underline, opacity\n\n**Responsive sweep:** Test at 3 viewport widths via browser MCP:\n- Desktop: 1440px\n- Tablet: 768px\n- Mobile: 390px\n- At each width, note which sections change layout (column → stack, sidebar disappears, etc.) and at approximately which breakpoint the change occurs.\n\nSave all findings to `docs/research/BEHAVIORS.md`. This is your behavior bible — reference it when writing every component spec.\n\n### Chrome MCP: precise motion & animation extraction\n\nUse **evaluate script** (or equivalent) in Chrome MCP **during** the scroll/click/hover sweeps so you capture **live computed values**, not guesses from DevTools screenshots alone.\n\n**Rules:**\n\n- **No hand-waving** — if the reference uses `0.45s cubic-bezier(0.4, 0, 0.2, 1)`, the spec and implementation use that exact string (or equivalent in Tailwind only when it truly matches).\n- **`@keyframes` source of truth** — for each animated element, record `getComputedStyle(el).animationName` and resolve the **full `@keyframes` rule** from stylesheets when allowed. Many sites inline CSS or use same-origin sheets: iterate `document.styleSheets` / `cssRules` inside `try/catch`; for **cross-origin** sheets that throw on `cssRules`, copy the rule from the **Network** response or **Sources** panel in Chrome and paste into `docs/research/MOTION.md` / the component spec. Never substitute a different easing or keyframe shape “because it looks close.”\n- **Scroll-driven animations** — for `animation-timeline: view()` / named timelines, capture **`view-timeline` / `scroll-timeline` on ancestors**, `animation-range`, `scroll-padding` / snap containers, and **which scroll container** is the timeline root (often `html`, sometimes a nested `overflow-y: auto` div).\n- **Before/after pairs** — for scroll- or hover-driven motion, capture computed `transform`, `opacity`, and longhand `animation-*` / `transition-*` **in both states** (see §7) and the **exact trigger** (px, ratio, or event).\n- **Libraries** — if Lenis/GSAP/Framer is present, document **version/hooks** from `package` on the page or script URLs, class hooks (e.g. `.lenis`), and any **inline style** the library sets at rest vs. mid-animation (sample via computed style after the interaction).\n- **Deliverable** — maintain `docs/research/MOTION.md`: page-level keyframes inventory, global scroll/smooth-scroll setup, and per-section pointers into component specs. Large pages may scope the audit script to a **section root selector** to avoid noise.\n\n**Motion audit script** (run via Chrome MCP; optional `rootSelector` limits to a subtree):\n\n```javascript\n(function motionAudit(rootSelector) {\n  const root = rootSelector ? document.querySelector(rootSelector) : document.body;\n  if (!root) return JSON.stringify({ error: 'root not found: ' + rootSelector });\n\n  function safeRules(sheet) {\n    try {\n      return [...sheet.cssRules];\n    } catch {\n      return [];\n    }\n  }\n\n  const walkKeyframes = (rules, out) => {\n    for (const rule of rules) {\n      if (rule.type === CSSRule.KEYFRAMES_RULE) {\n        out.push({ name: rule.name, cssText: rule.cssText });\n      } else if (rule.type === CSSRule.MEDIA_RULE && rule.cssRules) {\n        walkKeyframes(rule.cssRules, out);\n      }\n    }\n  };\n\n  const keyframes = [];\n  for (const sheet of document.styleSheets) {\n    walkKeyframes(safeRules(sheet), keyframes);\n  }\n\n  const animated = [];\n  const scope = root.querySelectorAll('*');\n  for (const el of scope) {\n    const cs = getComputedStyle(el);\n    const animName = cs.animationName;\n    const hasAnim = animName && animName !== 'none';\n    const transDurs = (cs.transitionDuration || '')\n      .split(',')\n      .map((s) => s.trim())\n      .filter(Boolean);\n    const hasTrans =\n      cs.transitionProperty &&\n      cs.transitionProperty !== 'none' &&\n      transDurs.some((d) => d !== '0s');\n    if (!hasAnim && !hasTrans) continue;\n\n    animated.push({\n      hint: el.tagName.toLowerCase() + (el.className\n        ? '.' + String(el.className).trim().split(/\\s+/).slice(0, 4).join('.')\n        : ''),\n      animation: cs.animation,\n      animationName: cs.animationName,\n      animationDuration: cs.animationDuration,\n      animationTimingFunction: cs.animationTimingFunction,\n      animationDelay: cs.animationDelay,\n      animationIterationCount: cs.animationIterationCount,\n      animationDirection: cs.animationDirection,\n      animationFillMode: cs.animationFillMode,\n      animationPlayState: cs.animationPlayState,\n      animationTimeline: cs.animationTimeline,\n      animationRange: cs.animationRange,\n      transition: cs.transition,\n      transitionProperty: cs.transitionProperty,\n      transitionDuration: cs.transitionDuration,\n      transitionTimingFunction: cs.transitionTimingFunction,\n      transitionDelay: cs.transitionDelay,\n      transform: cs.transform,\n      transformOrigin: cs.transformOrigin,\n      opacity: cs.opacity,\n      willChange: cs.willChange,\n      viewTimelineName: cs.viewTimelineName,\n      scrollTimelineName: cs.scrollTimelineName\n    });\n  }\n\n  return JSON.stringify(\n    {\n      prefersReducedMotion: window.matchMedia('(prefers-reduced-motion: reduce)').matches,\n      keyframes,\n      animatedElements: animated\n    },\n    null,\n    2\n  );\n})(''); // Pass section selector string, e.g. 'section.hero', instead of ''\n```\n\n### Page Topology\nMap out every distinct section of the page from top to bottom. Give each a working name. Document:\n- Their visual order\n- Which are fixed/sticky overlays vs. flow content\n- The overall page layout (scroll container, column structure, z-index layers)\n- Dependencies between sections (e.g., a floating nav that overlays everything)\n- **The interaction model** of each section (static, click-driven, scroll-driven, time-driven)\n\nSave this as `docs/research/PAGE_TOPOLOGY.md` — it becomes your assembly blueprint.\n\n## Phase 2: Foundation Build\n\nThis is sequential. Do it yourself (not delegated to an agent) since it touches many files:\n\n1. **Update fonts** in `layout.tsx` to match the target site's actual fonts\n2. **Update globals.css** with the target's color tokens, spacing values, keyframe animations, utility classes, and any **global scroll behaviors** (Lenis, smooth scroll CSS, scroll-snap on body)\n3. **Create TypeScript interfaces** in `src/types/` for the content structures you've observed\n4. **Extract SVG icons** — find all inline `<svg>` elements on the page, deduplicate them, and save as named React components in `src/components/icons.tsx`. Name them by visual function (e.g., `SearchIcon`, `ArrowRightIcon`, `LogoIcon`).\n5. **Download global assets** — write and run a Node.js script (`scripts/download-assets.mjs`) that downloads all images, videos, and other binary assets from the page to `public/`. Preserve meaningful directory structure.\n6. Verify: `npm run build` passes\n\n### Asset Discovery Script Pattern\n\nUse browser MCP to enumerate all assets on the page:\n\n```javascript\n// Run this via browser MCP to discover all assets\nJSON.stringify({\n  images: [...document.querySelectorAll('img')].map(img => ({\n    src: img.src || img.currentSrc,\n    alt: img.alt,\n    width: img.naturalWidth,\n    height: img.naturalHeight,\n    // Include parent info to detect layered compositions\n    parentClasses: img.parentElement?.className,\n    siblings: img.parentElement ? [...img.parentElement.querySelectorAll('img')].length : 0,\n    position: getComputedStyle(img).position,\n    zIndex: getComputedStyle(img).zIndex\n  })),\n  videos: [...document.querySelectorAll('video')].map(v => ({\n    src: v.src || v.querySelector('source')?.src,\n    poster: v.poster,\n    autoplay: v.autoplay,\n    loop: v.loop,\n    muted: v.muted\n  })),\n  backgroundImages: [...document.querySelectorAll('*')].filter(el => {\n    const bg = getComputedStyle(el).backgroundImage;\n    return bg && bg !== 'none';\n  }).map(el => ({\n    url: getComputedStyle(el).backgroundImage,\n    element: el.tagName + '.' + el.className?.split(' ')[0]\n  })),\n  svgCount: document.querySelectorAll('svg').length,\n  fonts: [...new Set([...document.querySelectorAll('*')].slice(0, 200).map(el => getComputedStyle(el).fontFamily))],\n  favicons: [...document.querySelectorAll('link[rel*=\"icon\"]')].map(l => ({ href: l.href, sizes: l.sizes?.toString() }))\n});\n```\n\nThen write a download script that fetches everything to `public/`. Use batched parallel downloads (4 at a time) with proper error handling.\n\n## Phase 3: Component Specification & Dispatch\n\nThis is the core loop. For each section in your page topology (top to bottom), you do THREE things: **extract**, **write the spec file**, then **dispatch builders**.\n\n### Step 1: Extract\n\nFor each section, use browser MCP to extract everything:\n\n1. **Screenshot** the section in isolation (scroll to it, screenshot the viewport). Save to `docs/design-references/`.\n\n2. **Extract CSS** for every element in the section. Use the extraction script below — don't hand-measure individual properties. Run it once per component container and capture the full output:\n\n```javascript\n// Per-component extraction — run via browser MCP\n// Replace SELECTOR with the actual CSS selector for the component\n(function(selector) {\n  const el = document.querySelector(selector);\n  if (!el) return JSON.stringify({ error: 'Element not found: ' + selector });\n  const props = [\n    'fontSize','fontWeight','fontFamily','lineHeight','letterSpacing','color',\n    'textTransform','textDecoration','backgroundColor','background',\n    'padding','paddingTop','paddingRight','paddingBottom','paddingLeft',\n    'margin','marginTop','marginRight','marginBottom','marginLeft',\n    'width','height','maxWidth','minWidth','maxHeight','minHeight',\n    'display','flexDirection','justifyContent','alignItems','gap',\n    'gridTemplateColumns','gridTemplateRows',\n    'borderRadius','border','borderTop','borderBottom','borderLeft','borderRight',\n    'boxShadow','overflow','overflowX','overflowY',\n    'position','top','right','bottom','left','zIndex',\n    'opacity','transform','transformOrigin','transition','transitionProperty','transitionDuration','transitionTimingFunction','transitionDelay',\n    'animation','animationName','animationDuration','animationTimingFunction','animationDelay','animationIterationCount','animationDirection','animationFillMode','animationPlayState','animationTimeline','animationRange','viewTimelineName','scrollTimelineName','cursor',\n    'objectFit','objectPosition','mixBlendMode','filter','backdropFilter',\n    'whiteSpace','textOverflow','WebkitLineClamp'\n  ];\n  function extractStyles(element) {\n    const cs = getComputedStyle(element);\n    const styles = {};\n    props.forEach(p => { const v = cs[p]; if (v && v !== 'none' && v !== 'normal' && v !== 'auto' && v !== '0px' && v !== 'rgba(0, 0, 0, 0)') styles[p] = v; });\n    return styles;\n  }\n  function walk(element, depth) {\n    if (depth > 4) return null;\n    const children = [...element.children];\n    return {\n      tag: element.tagName.toLowerCase(),\n      classes: element.className?.toString().split(' ').slice(0, 5).join(' '),\n      text: element.childNodes.length === 1 && element.childNodes[0].nodeType === 3 ? element.textContent.trim().slice(0, 200) : null,\n      styles: extractStyles(element),\n      images: element.tagName === 'IMG' ? { src: element.src, alt: element.alt, naturalWidth: element.naturalWidth, naturalHeight: element.naturalHeight } : null,\n      childCount: children.length,\n      children: children.slice(0, 20).map(c => walk(c, depth + 1)).filter(Boolean)\n    };\n  }\n  return JSON.stringify(walk(el, 0), null, 2);\n})('SELECTOR');\n```\n\n3. **Extract multi-state styles** — for any element with multiple states (scroll-triggered, hover, active tab), capture BOTH states:\n\n```javascript\n// State A: capture styles at current state (e.g., scroll position 0)\n// Then trigger the state change (scroll, click, hover via browser MCP)\n// State B: re-run the extraction script on the same element\n// The diff between A and B IS the behavior specification\n```\n\nRecord the diff explicitly: \"Property X changes from VALUE_A to VALUE_B, triggered by TRIGGER, with transition: TRANSITION_CSS.\"\n\n4. **Extract real content** — all text, alt attributes, aria labels, placeholder text. Use `element.textContent` for each text node. For tabbed/stateful content, **click each tab and extract content per state**.\n\n5. **Identify assets** this section uses — which downloaded images/videos from `public/`, which icon components from `icons.tsx`. Check for **layered images** (multiple `<img>` or background-images stacked in the same container).\n\n6. **Assess complexity** — how many distinct sub-components does this section contain? A distinct sub-component is an element with its own unique styling, structure, and behavior (e.g., a card, a nav item, a search panel).\n\n### Step 2: Write the Component Spec File\n\nFor each section (or sub-component, if you're breaking it up), create a spec file in `docs/research/components/`. This is NOT optional — every builder must have a corresponding spec file.\n\n**File path:** `docs/research/components/<component-name>.spec.md`\n\n**Template:**\n\n```markdown\n# <ComponentName> Specification\n\n## Overview\n- **Target file:** `src/components/<ComponentName>.tsx`\n- **Screenshot:** `docs/design-references/<screenshot-name>.png`\n- **Interaction model:** <static | click-driven | scroll-driven | time-driven>\n\n## DOM Structure\n<Describe the element hierarchy — what contains what>\n\n## Computed Styles (exact values from getComputedStyle)\n\n### Container\n- display: ...\n- padding: ...\n- maxWidth: ...\n- (every relevant property with exact values)\n\n### <Child element 1>\n- fontSize: ...\n- color: ...\n- (every relevant property)\n\n### <Child element N>\n...\n\n## Motion & animation (exact — from Chrome MCP `getComputedStyle` + stylesheet rules)\n\n- **Keyframe names in use:** ...\n- **Full `@keyframes` blocks** (paste `cssText` or equivalent); note **cross-origin** gaps and how you filled them\n- **Per-element:** shorthand/longhand `animation` and `transition` values **verbatim** for resting state\n- **Scroll/view timelines:** timeline name, range, scroll container, snap/observer thresholds\n- **Libraries:** Lenis / GSAP / Framer / Lottie — hooks, init pattern, timing copied from reference\n- **`prefers-reduced-motion`:** matched behavior on the reference\n\n## States & Behaviors\n\n### <Behavior name, e.g., \"Scroll-triggered floating mode\">\n- **Trigger:** <exact mechanism — scroll position 50px, IntersectionObserver rootMargin \"-30% 0px\", click on .tab-button, hover>\n- **State A (before):** maxWidth: 100vw, boxShadow: none, borderRadius: 0\n- **State B (after):** maxWidth: 1200px, boxShadow: 0 4px 20px rgba(0,0,0,0.1), borderRadius: 16px\n- **Transition:** transition: all 0.3s ease\n- **Implementation approach:** <CSS transition + scroll listener | IntersectionObserver | CSS animation-timeline | etc.>\n\n### Hover states\n- **<Element>:** <property>: <before> → <after>, transition: <value>\n\n## Per-State Content (if applicable)\n\n### State: \"Featured\"\n- Title: \"...\"\n- Subtitle: \"...\"\n- Cards: [{ title, description, image, link }, ...]\n\n### State: \"Productivity\"\n- Title: \"...\"\n- Cards: [...]\n\n## Assets\n- Background image: `public/images/<file>.webp`\n- Overlay image: `public/images/<file>.png`\n- Icons used: <ArrowIcon>, <SearchIcon> from icons.tsx\n\n## Text Content (verbatim)\n<All text content, copy-pasted from the live site>\n\n## Responsive Behavior\n- **Desktop (1440px):** <layout description>\n- **Tablet (768px):** <what changes — e.g., \"maintains 2-column, gap reduces to 16px\">\n- **Mobile (390px):** <what changes — e.g., \"stacks to single column, images full-width\">\n- **Breakpoint:** layout switches at ~<N>px\n```\n\nFill every section. If a section doesn't apply (e.g., no states for a static footer), write \"N/A\" — but think twice before marking States & Behaviors as N/A. Even a footer might have hover states on links.\n\n### Step 3: Dispatch Builders\n\nBased on complexity, dispatch builder agent(s) in worktree(s):\n\n**Simple section** (1-2 sub-components): One builder agent gets the entire section.\n\n**Complex section** (3+ distinct sub-components): Break it up. One agent per sub-component, plus one agent for the section wrapper that imports them. Sub-component builders go first since the wrapper depends on them.\n\n**What every builder agent receives:**\n- The full contents of its component spec file (inline in the prompt — don't say \"go read the spec file\")\n- Path to the section screenshot in `docs/design-references/`\n- Which shared components to import (`icons.tsx`, `cn()`, shadcn primitives)\n- The target file path (e.g., `src/components/HeroSection.tsx`)\n- Instruction to verify with `npx tsc --noEmit` before finishing\n- For responsive behavior: the specific breakpoint values and what changes\n\n**Don't wait.** As soon as you've dispatched the builder(s) for one section, move to extracting the next section. Builders work in parallel in their worktrees while you continue extraction.\n\n### Step 4: Merge\n\nAs builder agents complete their work:\n- Merge their worktree branches into main\n- You have full context on what each agent built, so resolve any conflicts intelligently\n- After each merge, verify the build still passes: `npm run build`\n- If a merge introduces type errors, fix them immediately\n\nThe extract → spec → dispatch → merge cycle continues until all sections are built.\n\n## Phase 4: Page Assembly\n\nAfter all sections are built and merged, wire everything together in `src/app/page.tsx`:\n\n- Import all section components\n- Implement the page-level layout from your topology doc (scroll containers, column structures, sticky positioning, z-index layering)\n- Connect real content to component props\n- Implement page-level behaviors: scroll snap, scroll-driven animations, dark-to-light transitions, intersection observers, smooth scroll (Lenis etc.)\n- Verify: `npm run build` passes clean\n\n## Phase 5: Visual QA Diff\n\nAfter assembly, do NOT declare the clone complete. Take side-by-side comparison screenshots:\n\n1. **Pixel-perfect comparison setup:** Match **Browser MCP: pixel-perfect capture** — same **CSS widths** (**1440**, **768**, **390**), **100% zoom**, and record `innerWidth`/`clientWidth`/`devicePixelRatio` on both reference and clone. Then open original and clone side-by-side or take paired screenshots at those widths.\n2. Compare section by section, top to bottom, at desktop (1440px)\n3. Compare at tablet (768px) and again at mobile (390px)\n4. For each discrepancy found:\n   - Check the component spec file — was the value extracted correctly?\n   - If the spec was wrong: re-extract from browser MCP, update the spec, fix the component\n   - If the spec was right but the builder got it wrong: fix the component to match the spec\n5. Test all interactive behaviors: scroll through the page, click every button/tab, hover over interactive elements\n6. **Motion QA (Chrome MCP again if needed):** For every animated or transitioned element, re-run `getComputedStyle` on the reference vs. the clone at the **same scroll position / interaction state** and compare **`animation-*`, `transition-*`, `transform`, `opacity`** longhands — durations and easings must match character-for-character where the cascade specifies them. Re-record any `@keyframes` the reference still exposes so the clone can be patched until timings align.\n7. Verify smooth scroll feels right, header transitions work, tab switching works, stagger and scroll-driven reveals match the reference\n\nOnly after this visual + motion QA pass is the clone complete.\n\n## Pre-Dispatch Checklist\n\nBefore dispatching ANY builder agent, verify you can check every box. If you can't, go back and extract more.\n\n- [ ] **Multi-pass DOM reconnaissance:** structure (`DOM_SKETCH.md`) + asset sweep + motion/flow sweeps + reconciliation; repeated at **768 / 390** when layout or DOM order differs from 1440\n- [ ] Spec file written to `docs/research/components/<name>.spec.md` with ALL sections filled\n- [ ] Every CSS value in the spec is from `getComputedStyle()`, not estimated — **verbatim**, including fractional px and color functions\n- [ ] **MCP pixel-perfect pass:** viewports **1440 / 768 / 390**, **100% zoom**, viewport diagnostic recorded; **`getBoundingClientRect()`** and **`::before`/`::after`** captured where they affect layout or visuals\n- [ ] Interaction model is identified and documented (static / click / scroll / time)\n- [ ] For stateful components: every state's content and styles are captured\n- [ ] For scroll-driven components: trigger threshold, before/after styles, and transition are recorded\n- [ ] For hover states: before/after values and transition timing are recorded\n- [ ] Motion: relevant `@keyframes` captured (or cross-origin gap documented with manual paste), `docs/research/MOTION.md` updated, component specs include **exact** `animation*` / `transition*` / timeline fields from Chrome MCP\n- [ ] All images in the section are identified (including overlays and layered compositions)\n- [ ] Responsive behavior is documented for **1440 / 768 / 390**\n- [ ] Text content is verbatim from the site, not paraphrased\n- [ ] The builder prompt is under ~150 lines of spec; if over, the section needs to be split\n\n## What NOT to Do\n\nThese are lessons from previous failed clones — each one cost hours of rework:\n\n- **Don't build click-based tabs when the original is scroll-driven (or vice versa).** Determine the interaction model FIRST by scrolling before clicking. This is the #1 most expensive mistake — it requires a complete rewrite, not a CSS fix.\n- **Don't extract only the default state.** If there are tabs showing \"Featured\" on load, click Productivity, Creative, Lifestyle and extract each one's cards/content. If the header changes on scroll, capture styles at position 0 AND position 100+.\n- **Don't miss overlay/layered images.** A background watercolor + foreground UI mockup = 2 images. Check every container's DOM tree for multiple `<img>` elements and positioned overlays.\n- **Don't build mockup components for content that's actually videos/animations.** Check if a section uses `<video>`, Lottie, or canvas before building elaborate HTML mockups of what the video shows.\n- **Don't approximate CSS classes.** \"It looks like `text-lg`\" is wrong if the computed value is `18px` and `text-lg` is `18px/28px` but the actual line-height is `24px`. Extract exact values.\n- **Don't build everything in one monolithic commit.** The whole point of this pipeline is incremental progress with verified builds at each step.\n- **Don't reference docs from builder prompts.** Each builder gets the CSS spec inline in its prompt — never \"see DESIGN_TOKENS.md for colors.\" The builder should have zero need to read external docs.\n- **Don't skip asset extraction.** Without real images, videos, and fonts, the clone will always look fake regardless of how perfect the CSS is.\n- **Don't give a builder agent too much scope.** If you're writing a builder prompt and it's getting long because the section is complex, that's a signal to break it into smaller tasks.\n- **Don't bundle unrelated sections into one agent.** A CTA section and a footer are different components with different designs — don't hand them both to one agent and hope for the best.\n- **Don't round or “clean up” extracted values** for Tailwind guesses: `12.8px` stays `12.8px` in the spec unless the clone uses an exactly equivalent value.\n- **Don't mix zoom levels or viewport widths** between reference extraction and QA — you will false-fail or false-pass pixel diffs.\n- **Don't skip responsive extraction.** If you only inspect at desktop width, the clone will break at tablet and mobile. Test at 1440, 768, and 390 during extraction.\n- **Don't forget smooth scroll libraries.** Check for Lenis (`.lenis` class), Locomotive Scroll, or similar. Default browser scrolling feels noticeably different and the user will spot it immediately.\n- **Don't dispatch builders without a spec file.** The spec file forces exhaustive extraction and creates an auditable artifact. Skipping it means the builder gets whatever you can fit in a prompt from memory.\n\n## Completion\n\nWhen done, report:\n- Multi-pass reconnaissance status (`docs/research/DOM_SKETCH.md` present; passes at **1440 / 768 / 390** as needed)\n- Total sections built\n- Total components created\n- Total spec files written (should match components)\n- Total assets downloaded (images, videos, SVGs, fonts)\n- Build status (`npm run build` result)\n- Visual QA results (any remaining discrepancies)\n- Any known gaps or limitations\n",
   "fileContext": [
     "AGENTS.md",
     "docs/research/**"

package/.augment/commands/launchframe.md

@@ -27,25 +27,23 @@ Before reconnaissance, write or update:
 
 After structure and styles match the reference, apply the **SaaS idea** to hero, headings, and primary CTAs where the reference uses interchangeable marketing copy, **without** changing layout grids, spacing, or motion from extracted specs. **Brand identity** (below) must be **original** for anything you ship as the user’s product — never pass off the reference company’s trademarks or distinctive marks.
 
-**Precedence:** For those surfaces (hero, main headings, primary CTAs), **final ship copy comes from the SaaS idea**, not the reference — still capture reference strings in research/specs if useful for QA diffs. Everywhere else, keep **verbatim** reference text unless the user overrides.
-
 ---
 
 ## Scope Defaults
 
 The target page(s) are the URL(s) you parsed in Pre-Flight. Clone exactly what's visible at each URL. Unless the user specifies otherwise, use these defaults:
 
-- **Fidelity level:** Pixel-perfect **layout, spacing, typography scale, and motion** (durations, easings, keyframes, triggers) matched to the reference. **Colors** match where you are emulating neutral/UI chrome from the reference; where **Brand identity** applies, use the **new palette** and assets while keeping the same CSS structure (fills land on different tokens). **Animations** stay reference-faithful unless the user opts out.
+- **Fidelity level:** Pixel-perfect — exact match in colors, spacing, typography, animations
 - **In scope:** Visual layout and styling, component structure and interactions, responsive design, mock data for demo purposes
 - **Out of scope:** Real backend / database, authentication, real-time features, SEO optimization, accessibility audit
 - **Customization:** Structure and visuals — pure emulation of the reference. **Marketing copy** — apply the parsed **SaaS idea** where hero/headlines/CTAs are interchangeable (see “SaaS copy overlay” above); do not invent a different product than the user’s idea.
-- **Brand identity:** The reference is a **pattern** for layout, motion, and UI craft — **not** permission to ship their brand. Unless the user **explicitly** asks for a faithful copy of the reference brand (e.g. licensed work, clearly labeled internal mock, private design audit), **invent an original brand** aligned with the SaaS idea: product name, wordmark or simple logomark (SVG or styled text) sized to the same logo slot, favicon / app icon, OG imagery, and a cohesive palette. Do **not** reuse their trademarked logo paths, mascot art, or distinctive illustrative brand assets; use originals or functional UI icons instead. Hero or lifestyle images that center the reference brand should be replaced with **original** imagery or neutral compositions that keep the **same layout rhythm**. **Keep** reference-faithful neutrals where they are clearly **non-brand** UI (borders, subtle grays, default body text color roles) unless they are unmistakably part of their distinctive palette. Note in `docs/research/LAUNCHFRAME.md` which marks and assets are **original brand** versus **layout-only** extraction.
+- **Brand identity:** The reference is a **pattern** for layout, motion, and UI craft — **not** permission to ship their brand. Unless the user **explicitly** asks for a faithful copy of the reference brand (e.g. licensed work, clearly labeled internal mock, private design audit), **invent an original brand** aligned with the SaaS idea: product name, wordmark or simple logomark (SVG or styled text) sized to the same logo slot, favicon / app icon, OG imagery, and a cohesive palette. Do **not** reuse their trademarked logo paths, mascot art, or distinctive illustrative brand assets; use originals or functional UI icons instead. Hero or lifestyle images that center the reference brand should be replaced with **original** imagery or neutral compositions that keep the **same layout rhythm**. Note in `docs/research/LAUNCHFRAME.md` which marks and assets are **original brand** versus **layout-only** extraction.
 
 If the user provides additional instructions (specific fidelity level, customizations, extra context), honor those over the defaults.
 
 ## Pre-Flight
 
-1. **Browser automation is required.** Check for available browser MCP tools (Chrome MCP, Playwright MCP, Browserbase MCP, Puppeteer MCP, etc.). Use whichever is available — if multiple exist, **prefer Chrome DevTools MCP** for inspection: it maps to the same engine the user sees and preserves **computed `animation` / `transition` / scroll-driven** values accurately. If none are detected, ask the user which browser tool they have and how to connect it. This skill cannot work without browser automation. **Motion parity:** every pass (reconnaissance, per-section extract, QA) must treat **motion as first-class** — same **durations, delays, easings (`cubic-bezier` / steps), iteration counts, fill modes, directions, keyframe percentages, and scroll/view-timeline bindings** as the live site, not “similar” motion.
+1. **Browser automation is required.** Check for available browser MCP tools (Chrome MCP, Playwright MCP, Browserbase MCP, Puppeteer MCP, etc.). Use whichever is available — if multiple exist, **prefer Chrome DevTools MCP** for inspection: it maps to the same engine the user sees and preserves **computed `animation` / `transition` / scroll-driven** values accurately. If none are detected, ask the user which browser tool they have and how to connect it. This skill cannot work without browser automation. **Pixel-perfect MCP:** every extract and QA pass must follow **Browser MCP: pixel-perfect capture** (Phase 1) — locked viewports, **verbatim** `getComputedStyle` strings (including fractional `px`, `rem`, `oklch`/`rgba`), and **geometry** from `getBoundingClientRect()` where layout matters. **Motion parity:** every pass must treat **motion as first-class** — same **durations, delays, easings (`cubic-bezier` / steps), iteration counts, fill modes, directions, keyframe percentages, and scroll/view-timeline bindings** as the live site, not “similar” motion.
 2. **Parse arguments** — extract every `http://` / `https://` URL token (there may be several). **SaaS idea** = the remaining non-URL text (trim outer quotes). Normalize and validate each URL; if any are invalid, or the SaaS idea is missing, ask the user once. For each valid URL, verify it is accessible via your browser MCP tool.
 3. Verify the base project builds: `npm run build`. The Next.js + shadcn/ui + Tailwind v4 scaffold should already be in place. If not, tell the user to set it up first.
 4. Create the output directories if they don't exist: `docs/research/`, `docs/research/components/`, `docs/design-references/`, `scripts/`. For multiple clones, also prepare per-site folders like `docs/research/<hostname>/` and `docs/design-references/<hostname>/`.
@@ -81,8 +79,6 @@ Look at each section and judge its complexity. A simple banner with a heading an
 
 Extract the actual text, images, videos, and SVGs from the live site. This is a clone, not a mockup. Use `element.textContent`, download every `<img>` and `<video>`, extract inline `<svg>` elements as React components. The only time you generate content is when something is clearly server-generated and unique per session.
 
-**Marketing surfaces** (hero, primary headings, main CTAs) follow the **SaaS copy overlay** and **Brand identity** rules — capture reference copy for audit, but **ship** the user’s idea there.
-
 **Prioritize** (see §0): downloadable imagery and backgrounds first, then SVG/icon layers, then motion. If you must **fabricate** an asset, prefer screenshot-based exports or traced vectors tied to measured box sizes — avoid unrelated stock art.
 
 **Layered assets matter.** A section that looks like one image is often multiple layers — a background watercolor/gradient, a foreground UI mockup PNG, an overlay icon. Inspect each container's full DOM tree and enumerate ALL `<img>` elements and background images within it, including absolutely-positioned overlays. Missing an overlay image makes the clone look empty even if the background is correct.
@@ -156,11 +152,48 @@ Navigate to the target URL with browser MCP.
 
 Follow **§0 (Visual crawl priority)** during the entire reconnaissance pass: images and backgrounds → SVGs/icons → motion/animations — before spending time on secondary copy tweaks.
 
+### Multi-pass DOM reconnaissance (mandatory)
+
+**A single scroll is not enough.** You must run the live page **multiple full cycles** in browser MCP — each time walking the **DOM**, the **asset graph**, and the **motion flow** — until what you write in specs is **exactly** what the reference does, not a remembered approximation.
+
+**Minimum passes** (run **Pass 1–4 at 1440px first**; **repeat Pass 1–3 at 768px and 390px** wherever layout or node order differs; reconciliation can be width-specific):
+
+1. **Structure pass (DOM, top → bottom):** Traverse from **`document.body`** in **depth-first order** (or section-by-section using regions you will later name in `PAGE_TOPOLOGY.md`). For every node that affects layout or visible content, record **tag**, **selector hint** (stable `id` / first meaningful classes), **`childNodes` order** (siblings matter — flex/grid gaps and wrappers are not optional), landmarks, and **which element is the scroll container** (`overflow` / `overscroll-behavior`). Produce a **literal hierarchy** in `docs/research/DOM_SKETCH.md` (whole page or per-section files under `docs/research/`) so builders cannot silently reorder or collapse wrapper divs. **Include “invisible” structure** (spacers, gradient overlays, absolutely positioned siblings).
+2. **Asset pass (images & layered media):** **Second full traversal** focused on §0 priority 1–2: every `<img>`, `<picture>` / `srcset`, `<video>` / `poster`, computed **`background-image`** on the element **and** ancestors, CSS masks, inline SVG / sprites. Cross-check with the Network panel. Every URL must end up **downloaded to `public/`** with a documented path, or marked **blocked / generated** with a substitute and reason.
+3. **Motion & flow pass:** **Third full journey** — slow scroll **top to bottom**, then **scroll again** (different speed) to catch lazy-mounted nodes, staggered `IntersectionObserver` reveals, and scroll-snapped stops. Run **`motionAudit`** (below) **per major section root** after scrolling that section into view. **Re-do click/hover sweeps** here while sampling `getComputedStyle` so **transitions and keyframe-driven flow** match the reference’s **order and timing**, not a generic fade-in.
+4. **Reconciliation pass:** **Fourth pass** — read `DOM_SKETCH.md`, `BEHAVIORS.md`, asset inventory, and `MOTION.md` together; list **gaps** (missing wrapper, second image layer, unnamed animation, wrong child order). For every gap, return to the site and **repeat Pass 1–3** for that subtree until the spec is closed.
+
+**Hard rule:** Every fact in a component spec must trace to **which pass** produced it. If you “kind of remember” a structure from an earlier visit, **run the page again** — do not guess.
+
 ### Screenshots
 - Take **full-page screenshots** at desktop (1440px) and mobile (390px) viewports
 - Save to `docs/design-references/` with descriptive names
 - These are your master reference — builders will receive section-specific crops/screenshots later
 
+### Browser MCP: pixel-perfect capture
+
+MCP-driven inspection must be **pixel-perfect**, not “close.” Apply this on **every** reconnaissance width and again in Phase 5.
+
+- **Viewport lock:** Resize the MCP-controlled window to exact **CSS widths** used for this project: **1440**, **768**, and **390** px (height large enough to avoid accidental mobile chrome quirks). After each resize, record **`window.innerWidth`**, **`document.documentElement.clientWidth`**, and **`devicePixelRatio`** (run the snippet below). Extraction and QA **use the same widths** so diffs are valid.
+- **Zoom:** Ensure **browser zoom is 100%** (not 90%/110%). Zoom breaks `px` parity and screenshots.
+- **Computed styles are verbatim:** When copying from `getComputedStyle()`, paste **exact returned strings** — including **fractional** lengths (`12.8px`), **negative** values, **`oklch()`/`lch()`/`rgba()`**, `letter-spacing`, `line-height`, `border`/`outline`, and **`calc()`**/`min()` that the cascade resolves to. **Do not round** for “clean” tokens (no `~16px`; no swapping `0.875rem` for `14px` unless the computed value truly equals that).
+- **Geometry:** When alignment, gutters, or sticky offsets matter, record **`getBoundingClientRect()`** (`x`, `y`, `width`, `height`) at the relevant scroll position and width — not only padding/margin strings. Fixed/sticky layers need **viewport-relative** positions.
+- **Pseudos & layers:** Extract **`::before` / `::after`** (and other pseudos the reference uses) with `getComputedStyle(el, '::before')` etc.; they often hold gradients, borders, or icons that define the pixel look.
+- **Anti-aliasing & text:** Prefer **real Chrome** via MCP so **font rasterization** matches; load the **same font files/weights** in Next.js as on the reference. Note `font-feature-settings` / `font-variation-settings` / `-webkit-font-smoothing` if non-default.
+- **Screenshots vs. DOM:** Screenshots are for **diffing**; the **spec values** still come from **computed style + rect**, not from eyeballing pixels.
+
+**Viewport diagnostic** (run via browser MCP after each resize):
+
+```javascript
+JSON.stringify({
+  innerWidth: window.innerWidth,
+  clientWidth: document.documentElement.clientWidth,
+  clientHeight: document.documentElement.clientHeight,
+  devicePixelRatio: window.devicePixelRatio,
+  scrollY: window.scrollY
+});
+```
+
 ### Global Extraction
 Extract these from the page before doing anything else:
 
@@ -198,8 +231,6 @@ This is a dedicated pass AFTER screenshots and BEFORE anything else. Its purpose
 - Mobile: 390px
 - At each width, note which sections change layout (column → stack, sidebar disappears, etc.) and at approximately which breakpoint the change occurs.
 
-Re-check **all three widths** again in **Phase 5: Visual QA Diff** so tablet regressions are not skipped.
-
 Save all findings to `docs/research/BEHAVIORS.md`. This is your behavior bible — reference it when writing every component spec.
 
 ### Chrome MCP: precise motion & animation extraction
@@ -208,11 +239,11 @@ Use **evaluate script** (or equivalent) in Chrome MCP **during** the scroll/clic
 
 **Rules:**
 
-- **No hand-waving** — if the reference uses `0.45s cubic-bezier(0.4, 0, 0.2, 1)`, the spec and implementation use that exact string (or Tailwind utilities **only** when the **resolved computed** animation/transition longhands match — class names alone are not proof).
+- **No hand-waving** — if the reference uses `0.45s cubic-bezier(0.4, 0, 0.2, 1)`, the spec and implementation use that exact string (or equivalent in Tailwind only when it truly matches).
 - **`@keyframes` source of truth** — for each animated element, record `getComputedStyle(el).animationName` and resolve the **full `@keyframes` rule** from stylesheets when allowed. Many sites inline CSS or use same-origin sheets: iterate `document.styleSheets` / `cssRules` inside `try/catch`; for **cross-origin** sheets that throw on `cssRules`, copy the rule from the **Network** response or **Sources** panel in Chrome and paste into `docs/research/MOTION.md` / the component spec. Never substitute a different easing or keyframe shape “because it looks close.”
-- **Scroll-driven animations** — for `animation-timeline: view()` / named timelines, capture **`view-timeline` / `scroll-timeline` on ancestors**, `animation-range`, `scroll-padding` / snap containers, and **which scroll container** is the timeline root (often `html`, sometimes a nested `overflow-y: auto` div). **`getComputedStyle` support varies** for scroll-animation longhands; also read **`cs.getPropertyValue('view-timeline-name')`**, **`scroll-timeline-name`**, **`animation-timeline`**, etc., when camelCase mirrors are missing in the engine you’re using.
+- **Scroll-driven animations** — for `animation-timeline: view()` / named timelines, capture **`view-timeline` / `scroll-timeline` on ancestors**, `animation-range`, `scroll-padding` / snap containers, and **which scroll container** is the timeline root (often `html`, sometimes a nested `overflow-y: auto` div).
 - **Before/after pairs** — for scroll- or hover-driven motion, capture computed `transform`, `opacity`, and longhand `animation-*` / `transition-*` **in both states** (see §7) and the **exact trigger** (px, ratio, or event).
-- **Libraries** — if Lenis/GSAP/Framer is present, document **version/hooks** from **script `src` URLs** (path often includes version), `//# sourceMappingURL` / comment banners in the bundle if visible, global identifiers on `window` when safe to read, **npm lockfile** only if this repo vendors that script, and **class/DOM hooks** (e.g. `.lenis`). Sample **computed style** at rest vs. mid-animation after interactions.
+- **Libraries** — if Lenis/GSAP/Framer is present, document **version/hooks** from `package` on the page or script URLs, class hooks (e.g. `.lenis`), and any **inline style** the library sets at rest vs. mid-animation (sample via computed style after the interaction).
 - **Deliverable** — maintain `docs/research/MOTION.md`: page-level keyframes inventory, global scroll/smooth-scroll setup, and per-section pointers into component specs. Large pages may scope the audit script to a **section root selector** to avoid noise.
 
 **Motion audit script** (run via Chrome MCP; optional `rootSelector` limits to a subtree):
@@ -246,8 +277,7 @@ Use **evaluate script** (or equivalent) in Chrome MCP **during** the scroll/clic
   }
 
   const animated = [];
-  // Include `root` — querySelectorAll('*') misses animations on the section container itself
-  const scope = [root, ...root.querySelectorAll('*')];
+  const scope = root.querySelectorAll('*');
   for (const el of scope) {
     const cs = getComputedStyle(el);
     const animName = cs.animationName;
@@ -259,7 +289,7 @@ Use **evaluate script** (or equivalent) in Chrome MCP **during** the scroll/clic
     const hasTrans =
       cs.transitionProperty &&
       cs.transitionProperty !== 'none' &&
-      transDurs.some((d) => d !== '0s' && d !== '0ms');
+      transDurs.some((d) => d !== '0s');
     if (!hasAnim && !hasTrans) continue;
 
     animated.push({
@@ -287,9 +317,7 @@ Use **evaluate script** (or equivalent) in Chrome MCP **during** the scroll/clic
       opacity: cs.opacity,
       willChange: cs.willChange,
       viewTimelineName: cs.viewTimelineName,
-      scrollTimelineName: cs.scrollTimelineName,
-      viewTimelineNameRaw: cs.getPropertyValue('view-timeline-name'),
-      scrollTimelineNameRaw: cs.getPropertyValue('scroll-timeline-name')
+      scrollTimelineName: cs.scrollTimelineName
     });
   }
 
@@ -376,7 +404,7 @@ For each section, use browser MCP to extract everything:
 
 1. **Screenshot** the section in isolation (scroll to it, screenshot the viewport). Save to `docs/design-references/`.
 
-2. **Extract CSS** for every element in the section. Use the extraction script below — don't hand-measure individual properties. Run it **once per spec scope**: usually the **section wrapper** selector (one subtree). For very large sections, run again on **inner selectors** (e.g. each card type) — the `depth` / `children.slice(0, 20)` limits are **deliberately capped** for MCP payload size; raise them or narrow `SELECTOR` when the spec is incomplete.
+2. **Extract CSS** for every element in the section. Use the extraction script below — don't hand-measure individual properties. Run it once per component container and capture the full output:
 
 ```javascript
 // Per-component extraction — run via browser MCP
@@ -403,7 +431,6 @@ For each section, use browser MCP to extract everything:
   function extractStyles(element) {
     const cs = getComputedStyle(element);
     const styles = {};
-    // Heuristic filter — for layout-audits, re-add any dropped `auto` / `normal` / transparent values the spec calls out as significant
     props.forEach(p => { const v = cs[p]; if (v && v !== 'none' && v !== 'normal' && v !== 'auto' && v !== '0px' && v !== 'rgba(0, 0, 0, 0)') styles[p] = v; });
     return styles;
   }
@@ -567,17 +594,16 @@ After all sections are built and merged, wire everything together in `src/app/pa
 
 After assembly, do NOT declare the clone complete. Take side-by-side comparison screenshots:
 
-1. Open the original site and your clone side-by-side (or take screenshots at the same viewport widths)
+1. **Pixel-perfect comparison setup:** Match **Browser MCP: pixel-perfect capture** — same **CSS widths** (**1440**, **768**, **390**), **100% zoom**, and record `innerWidth`/`clientWidth`/`devicePixelRatio` on both reference and clone. Then open original and clone side-by-side or take paired screenshots at those widths.
 2. Compare section by section, top to bottom, at desktop (1440px)
-3. Compare again at tablet (768px) — same widths as **Responsive sweep**
-4. Compare again at mobile (390px)
-5. For each discrepancy found:
+3. Compare at tablet (768px) and again at mobile (390px)
+4. For each discrepancy found:
    - Check the component spec file — was the value extracted correctly?
    - If the spec was wrong: re-extract from browser MCP, update the spec, fix the component
    - If the spec was right but the builder got it wrong: fix the component to match the spec
-6. Test all interactive behaviors: scroll through the page, click every button/tab, hover over interactive elements
-7. **Motion QA (Chrome MCP again if needed):** For every animated or transitioned element, re-run `getComputedStyle` on the reference vs. the clone at the **same scroll position / interaction state** and compare **`animation-*`, `transition-*`, `transform`, `opacity`** longhands — **resolved** durations/easings must match where the cascade specifies them (Tailwind classes are fine when they compile to the same longhands; verify in **Computed**, not by class name alone). Re-record any `@keyframes` the reference still exposes so the clone can be patched until timings align.
-8. Verify smooth scroll feels right, header transitions work, tab switching works, stagger and scroll-driven reveals match the reference
+5. Test all interactive behaviors: scroll through the page, click every button/tab, hover over interactive elements
+6. **Motion QA (Chrome MCP again if needed):** For every animated or transitioned element, re-run `getComputedStyle` on the reference vs. the clone at the **same scroll position / interaction state** and compare **`animation-*`, `transition-*`, `transform`, `opacity`** longhands — durations and easings must match character-for-character where the cascade specifies them. Re-record any `@keyframes` the reference still exposes so the clone can be patched until timings align.
+7. Verify smooth scroll feels right, header transitions work, tab switching works, stagger and scroll-driven reveals match the reference
 
 Only after this visual + motion QA pass is the clone complete.
 
@@ -585,16 +611,18 @@ Only after this visual + motion QA pass is the clone complete.
 
 Before dispatching ANY builder agent, verify you can check every box. If you can't, go back and extract more.
 
+- [ ] **Multi-pass DOM reconnaissance:** structure (`DOM_SKETCH.md`) + asset sweep + motion/flow sweeps + reconciliation; repeated at **768 / 390** when layout or DOM order differs from 1440
 - [ ] Spec file written to `docs/research/components/<name>.spec.md` with ALL sections filled
-- [ ] Every CSS value in the spec is from `getComputedStyle()`, not estimated
+- [ ] Every CSS value in the spec is from `getComputedStyle()`, not estimated — **verbatim**, including fractional px and color functions
+- [ ] **MCP pixel-perfect pass:** viewports **1440 / 768 / 390**, **100% zoom**, viewport diagnostic recorded; **`getBoundingClientRect()`** and **`::before`/`::after`** captured where they affect layout or visuals
 - [ ] Interaction model is identified and documented (static / click / scroll / time)
 - [ ] For stateful components: every state's content and styles are captured
 - [ ] For scroll-driven components: trigger threshold, before/after styles, and transition are recorded
 - [ ] For hover states: before/after values and transition timing are recorded
 - [ ] Motion: relevant `@keyframes` captured (or cross-origin gap documented with manual paste), `docs/research/MOTION.md` updated, component specs include **exact** `animation*` / `transition*` / timeline fields from Chrome MCP
 - [ ] All images in the section are identified (including overlays and layered compositions)
-- [ ] Responsive behavior is documented for desktop, tablet, and mobile (1440 / 768 / 390)
-- [ ] Text: **verbatim** from the reference **except** hero, main headings, and primary CTAs — those match the **SaaS idea** (reference copy still captured for audit if useful)
+- [ ] Responsive behavior is documented for **1440 / 768 / 390**
+- [ ] Text content is verbatim from the site, not paraphrased
 - [ ] The builder prompt is under ~150 lines of spec; if over, the section needs to be split
 
 ## What NOT to Do
@@ -611,6 +639,8 @@ These are lessons from previous failed clones — each one cost hours of rework:
 - **Don't skip asset extraction.** Without real images, videos, and fonts, the clone will always look fake regardless of how perfect the CSS is.
 - **Don't give a builder agent too much scope.** If you're writing a builder prompt and it's getting long because the section is complex, that's a signal to break it into smaller tasks.
 - **Don't bundle unrelated sections into one agent.** A CTA section and a footer are different components with different designs — don't hand them both to one agent and hope for the best.
+- **Don't round or “clean up” extracted values** for Tailwind guesses: `12.8px` stays `12.8px` in the spec unless the clone uses an exactly equivalent value.
+- **Don't mix zoom levels or viewport widths** between reference extraction and QA — you will false-fail or false-pass pixel diffs.
 - **Don't skip responsive extraction.** If you only inspect at desktop width, the clone will break at tablet and mobile. Test at 1440, 768, and 390 during extraction.
 - **Don't forget smooth scroll libraries.** Check for Lenis (`.lenis` class), Locomotive Scroll, or similar. Default browser scrolling feels noticeably different and the user will spot it immediately.
 - **Don't dispatch builders without a spec file.** The spec file forces exhaustive extraction and creates an auditable artifact. Skipping it means the builder gets whatever you can fit in a prompt from memory.
@@ -618,6 +648,7 @@ These are lessons from previous failed clones — each one cost hours of rework:
 ## Completion
 
 When done, report:
+- Multi-pass reconnaissance status (`docs/research/DOM_SKETCH.md` present; passes at **1440 / 768 / 390** as needed)
 - Total sections built
 - Total components created
 - Total spec files written (should match components)