From ecd9ec77aa5799c3e59b23b36eb1d633477a825e Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Mon, 28 Sep 2026 22:58:46 +0100
Subject: [PATCH 1/9] fix: keep cinematic opening frame visible

---
 .../funnels/ScrollVideoBackground.tsx         |  8 +++--
 .../cinematic-marketing-experience.spec.ts    | 35 +++++++++++++++++++
 2 files changed, 40 insertions(+), 3 deletions(-)

diff --git a/src/components/funnels/ScrollVideoBackground.tsx b/src/components/funnels/ScrollVideoBackground.tsx
index 1cbbc19..2815a70 100644
--- a/src/components/funnels/ScrollVideoBackground.tsx
+++ b/src/components/funnels/ScrollVideoBackground.tsx
@@ -9,6 +9,7 @@ const BOOTSTRAP_SPRITE_URL = '/funnels/cinematic-bootstrap.jpg'
 const BOOTSTRAP_FRAME_COUNT = 24
 const BOOTSTRAP_COLUMNS = 6
 const BOOTSTRAP_ROWS = 4
+const FIRST_FRAME_TIME_SECONDS = 0.04
 
 type ScrollVideoBackgroundProps = {
   containerRef?: RefObject<HTMLElement | null>
@@ -74,13 +75,14 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
     const image = new Image()
     image.decoding = 'async'
     image.fetchPriority = 'high'
-    image.src = BOOTSTRAP_SPRITE_URL
     image.onload = () => {
       bootstrapImageRef.current = image
       setBootstrapReady(true)
     }
+    image.src = BOOTSTRAP_SPRITE_URL
     return () => {
       image.onload = null
+      image.onerror = null
       bootstrapImageRef.current = null
     }
   }, [])
@@ -175,7 +177,7 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
 
       for (const index of buildExtractionOrder(count)) {
         if (cancelled) break
-        const time = (index / Math.max(1, count - 1)) * Math.max(0, duration - 0.05)
+        const time = Math.max(FIRST_FRAME_TIME_SECONDS, (index / Math.max(1, count - 1)) * Math.max(0, duration - 0.05))
         await seekTo(time)
         if (cancelled) break
         ctx.drawImage(source, 0, 0, width, height)
@@ -282,5 +284,5 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
   }, [bootstrapReady, cacheReady, videoReady])
 
   const canvasVisible = bootstrapReady || cacheReady
-  return <><link rel="preload" as="image" href={BOOTSTRAP_SPRITE_URL} /><div className={`cm-video ${className}`.trim()} aria-hidden="true"><div className={`cm-video__poster ${bootstrapReady || videoReady || cacheReady ? 'is-hidden' : ''}`} /><video ref={videoRef} className={`cm-video__element ${canvasVisible || cacheReady ? 'is-hidden' : videoReady ? 'is-visible' : ''}`} src={VIDEO_URL} muted playsInline preload="metadata" /><canvas ref={canvasRef} className={`cm-video__canvas ${canvasVisible ? 'is-visible' : ''}`} /><div className="cm-video__veil" /></div></>
+  return <><link rel="preload" as="image" href={BOOTSTRAP_SPRITE_URL} /><div className={`cm-video ${className}`.trim()} aria-hidden="true"><div className={`cm-video__poster ${bootstrapReady || videoReady || cacheReady ? 'is-hidden' : ''}`} /><video ref={videoRef} className={`cm-video__element ${canvasVisible || cacheReady ? 'is-hidden' : videoReady ? 'is-visible' : ''}`} src={VIDEO_URL} muted playsInline preload="metadata" /><canvas ref={canvasRef} className={`cm-video__canvas ${canvasVisible ? 'is-visible' : ''}`} data-frame-cache-ready={cacheReady ? 'true' : 'false'} /><div className="cm-video__veil" /></div></>
 }
diff --git a/tests/evidence/cinematic-marketing-experience.spec.ts b/tests/evidence/cinematic-marketing-experience.spec.ts
index 5054a7d..68741b3 100644
--- a/tests/evidence/cinematic-marketing-experience.spec.ts
+++ b/tests/evidence/cinematic-marketing-experience.spec.ts
@@ -274,6 +274,41 @@ test.describe('cinematic marketing experience', () => {
     })
   }
 
+  test('keeps the opening cinematic frame painted when the video cache takes over', async ({ page }) => {
+    await page.setViewportSize({ width: 1440, height: 900 })
+    await page.emulateMedia({ reducedMotion: 'no-preference' })
+    await page.goto(url('/'), { waitUntil: 'domcontentloaded' })
+
+    const canDecodeVideo = await page.evaluate(() => Boolean(
+      document.createElement('video').canPlayType('video/mp4; codecs="avc1.640028"'),
+    ))
+    const canvas = page.locator('.cm-video__canvas')
+    await expect(canvas).toHaveClass(/is-visible/)
+    if (canDecodeVideo) {
+      await expect(canvas).toHaveAttribute('data-frame-cache-ready', 'true', { timeout: 20000 })
+    } else {
+      await page.waitForTimeout(4500)
+    }
+
+    const paintedPixels = await canvas.evaluate(element => {
+      const canvas = element as HTMLCanvasElement
+      const sample = document.createElement('canvas')
+      sample.width = 32
+      sample.height = 18
+      const context = sample.getContext('2d')
+      if (!context) return 0
+      context.drawImage(canvas, 0, 0, sample.width, sample.height)
+      const pixels = context.getImageData(0, 0, sample.width, sample.height).data
+      let count = 0
+      for (let index = 3; index < pixels.length; index += 4) {
+        if (pixels[index] > 0) count += 1
+      }
+      return count
+    })
+
+    expect(paintedPixels, 'the opening frame should not be replaced by a transparent video frame').toBeGreaterThan(0)
+  })
+
   test('reduced motion keeps all content readable without canvas scrubbing', async ({ page }) => {
     await page.setViewportSize({ width: 1440, height: 900 })
     await page.emulateMedia({ reducedMotion: 'reduce' })

From 3d34d3c688433134a157f1297c24f6afbd30fb47 Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Tue, 29 Sep 2026 09:42:59 +0100
Subject: [PATCH 2/9] fix: finalize cinematic ProChat marketing experience

---
 AGENTS.md                                     |  25 +-
 PRODUCT.md                                    | 149 +-----
 README.md                                     |  35 +-
 docs/deployment.md                            |  10 +-
 docs/design/MOTION_STORYBOARD.md              |   2 +
 docs/homepage-validation-plan.md              | 458 ++----------------
 docs/implementation-plan.md                   | 121 ++---
 docs/overview.md                              |  14 +-
 docs/platform/PERFORMANCE_STRATEGY.md         |  19 +-
 docs/product/PUBLIC_PAGE_ARCHITECTURE.md      |   4 +-
 docs/product/agent-mode-progress.md           |   4 +-
 docs/repo-status.md                           |  81 +---
 docs/roadmap.md                               |  90 ++--
 docs/strategy.md                              |  45 +-
 src/components/CinematicMarketingPage.tsx     |   1 -
 src/components/CinematicMarketingShell.tsx    |   2 +-
 .../cinematic/cinematic-marketing.css         |   4 +-
 src/components/cinematic/cinematic-media.css  |   7 +-
 .../cinematic/cinematic-reference-final.css   |   7 +-
 .../funnels/ScrollVideoBackground.tsx         | 146 ++++--
 .../evidence/cinematic-final-quality.spec.ts  | 211 ++++++++
 .../cinematic-marketing-experience.spec.ts    |   3 +-
 22 files changed, 546 insertions(+), 892 deletions(-)
 create mode 100644 tests/evidence/cinematic-final-quality.spec.ts

diff --git a/AGENTS.md b/AGENTS.md
index edbc2c1..5e62d75 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -10,16 +10,17 @@ Current product hierarchy:
 
 ```text
 ProChat
-├── ProChat Memory
-│   └── ProChat Memory for QA
-└── ProChat Workbench
+├── Infinite Brain
+│   ├── Evermind — free, open-source, human-owned memory
+│   └── Nevermind — paid capability and convenience layer
+└── Mastermind — separate free, open-source planning/orchestration product
 ```
 
 Current positioning:
 
-- **ProChat Memory** is the flagship product.
-- **ProChat Memory for QA** is the first launch niche and first discipline-specific edition of ProChat Memory.
-- **ProChat Workbench** is the second product and should be presented separately from Memory.
+- Evermind and Nevermind are the primary connected ProChat product system; Mastermind is separate.
+- ProChat Memory for QA is paused/historical, not a current public product.
+- Mastermind is the current public name for the former Workbench product line.
 - ProChat OS, SaaSKit, ProKit, UXKit, WaaSKit, MikeOSS, and similar older names are legacy, historical, archived, external, or internal references unless Mind explicitly reclassifies them.
 - ProChat Answers, ProChat Automations, API access, and MCP integrations are capabilities, use cases, or future interfaces, not current products.
 - BuildFlow may appear only as a technical/internal compatibility identifier for Workbench where required.
@@ -29,10 +30,10 @@ Current positioning:
 Mind controls company and product strategy:
 
 ```text
-mind/wiki/organisations/prochat/README.md
-mind/wiki/organisations/prochat/brand/
-mind/wiki/organisations/prochat/legal/
-mind/wiki/organisations/prochat/growth/
+mind/organizations/prochat/README.md
+mind/organizations/prochat/brand/
+mind/organizations/prochat/legal/
+mind/organizations/prochat/growth/
 ```
 
 This repository controls website implementation facts: routes, components, content rendering, analytics hooks, environment contracts, docs automation, deployment behavior, and code-facing architecture.
@@ -48,7 +49,7 @@ Use these docs for implementation and website execution context only:
 - `docs/strategy.md` — repository-local website strategy subordinate to Mind;
 - `docs/roadmap.md` — website implementation roadmap subordinate to Mind;
 - `docs/implementation-plan.md` — website implementation plan subordinate to Mind;
-- `docs/prochat-memory-website-design-strategy.md` — design brief for the Memory-led website;
+- `docs/product/PUBLIC_PAGE_ARCHITECTURE.md` — historical/superseded page plan; not current route or product authority;
 - `docs/website-copy-blueprint.md` — website copy guidance subordinate to Mind;
 - `DESIGN.md` — visual/design-system guidance only.
 
@@ -84,7 +85,7 @@ The environment variable contract lives in `docs-public/environment.md` and `.en
 4. Never adjust environment contracts without updating `docs-public/environment.md` and `.env.example` where relevant.
 5. Do not reintroduce legacy systems that were removed unless the task explicitly asks for a historical archive or redirect.
 6. Node 20 is the supported runtime for Docker, CI, and local tooling unless the repo contract changes.
-7. Stripe uses mode-based environment keys (`STRIPE_MODE`, `STRIPE_SECRET_KEY_{TEST|LIVE}`, etc.).
+7. Do not restore retired Stripe runtime configuration; current environment requirements are documented in `docs-public/environment.md` and `.env.example`.
 8. Documentation integrity checks (`scripts/check-env-docs.js`, `scripts/check-doc-links.js`, `npm run docs:validate`) run automatically in CI.
 
 ## Project principles
diff --git a/PRODUCT.md b/PRODUCT.md
index 9b8326a..932cdf2 100644
--- a/PRODUCT.md
+++ b/PRODUCT.md
@@ -1,144 +1,41 @@
 # ProChat Product Context
 
-**Status:** repository-local implementation context  
-**Canonical strategy:** Mind  
+**Status:** repository-local implementation context
+**Canonical product strategy:** Mind, `mind/organizations/prochat/brand/`
 **Applies to:** ProChat website design, implementation, review, and polish
 
-This file gives design and implementation agents the minimum product context required to work safely in this repository. It does not replace canonical Mind strategy.
+This file summarizes canonical product facts for website implementation. It must not replace Mind or invent product hierarchy, pricing, maturity, or claims.
 
-## Company
-
-ProChat is a memory-first software company.
-
-It builds local products that keep project knowledge reusable and connect familiar AI interfaces to real project work.
-
-## Current product family
-
-ProChat has exactly two current products:
-
-1. **ProChat Memory** — the flagship product for reusable project knowledge.
-2. **ProChat Workbench** — a ChatGPT-first local builder workbench.
-
-**ProChat Memory for QA** is the first discipline-specific edition of ProChat Memory. It is not a third product.
-
-Future API access, automation, and MCP integrations are capabilities or interfaces within current products, not separate products.
-
-## Flagship product
-
-ProChat Memory is a local, Markdown-first project memory system.
-
-It helps people capture, review, retrieve, correct, improve, and retire reusable project knowledge without requiring ProChat to host customer memory.
-
-The trust workflow is:
+## Current product system
 
 ```text
-current evidence
-→ draft lesson
-→ sanitization and scope
-→ human review
-→ approved memory
-→ relevant retrieval
-→ correction or retirement
-```
-
-Current evidence and human judgment remain stronger than stored memory.
-
-## Current launch focus
-
-The first and only current discipline-specific edition is **ProChat Memory for QA**.
-
-The current primary user is an individual QA tester working with repeated failures, investigations, evidence, fixes, selectors, environments, test data, and review decisions.
-
-The selected QA beta is:
+ProChat → Infinite Brain → Evermind ↔ Nevermind
+                         Mind       Brain
 
-- free only for approved beta participants;
-- manually selected;
-- public source-available, not open source;
-- without a fixed end date;
-- in active development;
-- focused on measured repeated value rather than broad sales.
-
-## Workbench
-
-ProChat Workbench connects ChatGPT reasoning to repositories, documentation, notes, and knowledge folders through controlled local operations.
-
-Its canonical sequence is:
-
-```text
-request
-→ exact local context
-→ guarded change
-→ targeted validation
-→ explicit Git action
+Mastermind — separate free, open-source planning/orchestration product
 ```
 
-Workbench must not be presented as an autonomous agent that silently controls a repository.
-
-## Homepage objective
-
-The homepage must make these points understandable:
-
-1. Useful project work should leave useful memory behind.
-2. ProChat Memory is the flagship.
-3. Memory becomes trusted through evidence, sanitization, scope, and human review.
-4. Customer memory remains local in the current product model.
-5. Relevant memory should enter a task without loading the entire archive.
-6. The currently available edition is ProChat Memory for QA.
-7. Workbench is the second product and shares the same controlled, inspectable philosophy.
-
-## Primary conversion
+- **Evermind** is free, open-source, durable, human-owned memory. It is local-first, readable, portable, inspectable, editable, provenance-aware, reviewed, and useful without Nevermind.
+- **Nevermind** is the paid capability and convenience layer around that memory: installation, configuration, retrieval, integrations, automation, validation, updates, and optional managed operations.
+- **Mastermind** is a separate free, open-source ChatGPT-native planner/orchestrator for bounded Codex execution. It replaces Workbench as the current public name.
+- **ProChat Memory for QA** is paused/on ice and historical. Keep its documentation and code as evidence, but do not present it as a current public product.
+- Mind and Brain remain distinct; do not introduce MindOS as another customer-facing brand.
 
-**Explore ProChat Memory for QA**
+## Public website contract
 
-Secondary conversions:
+Canonical routes are `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, `/contact`, `/privacy`, and `/terms`. This cinematic visual contract applies to the first four routes only. The remaining public routes retain their existing page-specific presentation and legal substance; do not redesign them as part of the cinematic release.
 
-- Join the selected QA beta
-- See how Memory works
-- Explore ProChat Workbench
+The home page product sentence is: “Evermind remembers. Nevermind brings context. Mastermind directs the work.” Preserve the distinct product roles and use accurate route-specific calls to action.
 
-## Brand character
+## Copy and claim boundaries
 
-The company should feel:
+- Do not claim zero hallucinations, guaranteed savings, universal compatibility, automatic trusted memory, or a hosted customer-memory platform without canonical evidence.
+- Do not imply future capabilities, APIs, MCP integrations, automation, or managed services are currently available unless verified in both Mind and implementation.
+- Do not present paused ProChat Memory for QA, Workbench, or BuildFlow as a current public product. BuildFlow may remain only as a technical/internal compatibility identifier where required.
+- Do not invent pricing, legal terms, encryption, support commitments, maturity, or customer-data handling claims.
 
-- reliable;
-- trustworthy;
-- stable;
-- clean;
-- minimal;
-- logical;
-- structural;
-- simple;
-- premium;
-- technically credible.
+## Design and implementation authority
 
-The website should feel like a calm working system for structured knowledge, not an AI spectacle.
+The active cinematic design system for the first four routes is implemented by `CinematicMarketingShell`, `CinematicMarketingPage`, `CinematicProductFunnel`, `CinematicMediaPage`, and `ScrollVideoBackground`. Keep route data distinct from shared page geometry. Product pages use scroll-driven media. Reduced-motion users must retain readable content and see one stable media layer. Treat other routes as regression-only scope in this release.
 
-## Copy boundaries
-
-Do not claim:
-
-- zero hallucinations;
-- automatic trusted memory;
-- guaranteed savings;
-- universal compatibility;
-- a hosted customer-memory platform;
-- encryption that is not documented;
-- that no information can ever reach an external AI provider;
-- finalized Memory pricing or enterprise support;
-- that future capabilities are current products.
-
-Use “durable” and “reusable” for Memory rather than making “persistent” the primary category language.
-
-Use “customer memory remains on the customer’s computer” rather than a blanket “private” claim.
-
-## Canonical references
-
-```text
-mind/wiki/organisations/prochat/brand/product-strategy.md
-mind/wiki/organisations/prochat/brand/product-naming-architecture.md
-mind/wiki/organisations/prochat/brand/product-roadmap.md
-mind/wiki/organisations/prochat/brand/canonical-homepage-copy.md
-mind/wiki/organisations/prochat/brand/global-design-foundation.md
-mind/wiki/organisations/prochat/brand/website-visual-motion-system.md
-mind/wiki/organisations/prochat/brand/website-build-contract.md
-```
+Consult current implementation documentation in `docs/overview.md`, `docs/strategy.md`, `docs/roadmap.md`, and `docs/implementation-plan.md`. `docs/product/PUBLIC_PAGE_ARCHITECTURE.md` and `docs/homepage-*` plans predate the current product/routes and must be treated as historical until explicitly reconciled.
diff --git a/README.md b/README.md
index 468ba78..99cd653 100644
--- a/README.md
+++ b/README.md
@@ -21,14 +21,14 @@ Mind is canonical for ProChat philosophy, product hierarchy, naming, positioning
 Before changing website positioning, product navigation, roadmap language, legal copy, growth claims, or marketing page structure, read:
 
 ```text
-mind/wiki/organisations/prochat/brand/README.md
-mind/wiki/organisations/prochat/brand/product-strategy.md
-mind/wiki/organisations/prochat/brand/product-naming-architecture.md
-mind/wiki/organisations/prochat/brand/product-roadmap.md
-mind/wiki/organisations/prochat/brand/canonical-homepage-copy.md
-mind/wiki/organisations/prochat/brand/global-design-foundation.md
-mind/wiki/organisations/prochat/brand/website-visual-motion-system.md
-mind/wiki/organisations/prochat/brand/website-build-contract.md
+mind/organizations/prochat/brand/README.md
+mind/organizations/prochat/brand/product-strategy.md
+mind/organizations/prochat/brand/product-naming-architecture.md
+mind/organizations/prochat/brand/product-roadmap.md
+mind/organizations/prochat/brand/canonical-homepage-copy.md
+mind/organizations/prochat/brand/global-design-foundation.md
+mind/organizations/prochat/brand/website-visual-motion-system.md
+mind/organizations/prochat/brand/website-build-contract.md
 ```
 
 This repository must not independently redefine ProChat philosophy, product hierarchy, naming, positioning, business stage, legal-policy direction, growth policy, or cross-product roadmap.
@@ -37,21 +37,22 @@ If local website implementation facts appear to conflict with Mind, report and r
 
 ## Current product boundary
 
-Mind currently defines exactly two ProChat products:
+Mind currently defines this product system:
 
 ```text
 ProChat
-├── ProChat Memory
-│   └── ProChat Memory for QA
-└── ProChat Workbench
+├── Infinite Brain
+│   ├── Evermind — free, open-source, human-owned memory
+│   └── Nevermind — paid capability and convenience layer
+└── Mastermind — separate free, open-source planning/orchestration product
 ```
 
 Current website implications:
 
-- ProChat Memory is the flagship product.
-- ProChat Memory for QA is the first launch niche and first discipline-specific edition.
-- ProChat Workbench is the second product and should be presented separately from Memory.
-- ProChat Answers, ProChat Automations, API access, and MCP integrations are capabilities or future interfaces, not current products.
+- Evermind and Nevermind are the primary connected product system; Mastermind is separate.
+- ProChat Memory for QA is paused/historical and is not in normal public navigation.
+- Mastermind is the current public name for the former Workbench line.
+- ProChat Answers, ProChat Automations, API access, and MCP integrations are capabilities or future interfaces, not current products unless Mind explicitly changes this.
 - ProChat OS, SaaSKit, ProKit, UXKit, WaaSKit, and MikeOSS are legacy, historical, external, or archived references where relevant; they are not current ProChat products.
 - BuildFlow may appear only as a technical/internal compatibility identifier for Workbench where required.
 
@@ -68,7 +69,7 @@ Use these docs for website implementation only:
 7. `docs/migration/` — foundational legacy sweep, migration matrix, and content, route, component, style, motion, asset, and dependency audits.
 8. `docs/homepage-design-spec.md` — central homepage design brief and closed decisions.
 9. `docs/homepage-visual-storyboard.md` and `docs/homepage-example-data.md` — visual states and sanitized prototype content.
-10. `docs/homepage-technical-design.md` and `docs/homepage-design-orchestration.md` — build architecture, Brain skill methods, tooling, and GPT-5.6 Sol workflow.
+10. `docs/homepage-technical-design.md` and `docs/homepage-design-orchestration.md` — historical build-planning evidence; check current source before relying on details.
 11. `docs/homepage-validation-plan.md` — visual, motion, accessibility, performance, and release gates.
 12. `docs/roadmap.md` — canonical 15-phase public-platform roadmap and continuous-governance status.
 13. `docs/implementation-plan.md` — dependency-gated implementation tasks with validation, rollback, and commit boundaries.
diff --git a/docs/deployment.md b/docs/deployment.md
index 19846e4..e72c87a 100644
--- a/docs/deployment.md
+++ b/docs/deployment.md
@@ -1,6 +1,6 @@
 # Deployment
 
-This document covers the current lean ProChat deployment path.
+This document covers the current ProChat deployment and verification path. Production releases flow through `.github/workflows/main.yml` on `main`; do not bypass its checks or deploy a local build directly.
 
 ## Production source
 
@@ -55,14 +55,14 @@ After rollout:
 3. verify intentional compatibility redirects;
 4. verify retired product routes do not render retired bodies;
 5. check production health/console evidence relevant to the change;
-6. for cinematic funnel releases, verify `/evermind` and `/nevermind` at desktop and mobile widths, including first-scroll behavior before the remote MP4 is ready.
+6. for cinematic releases, verify `/`, `/evermind`, `/nevermind`, and `/mastermind` at desktop and mobile widths, including first-scroll and reverse-scroll behavior, reduced motion, and production media readiness.
 
 Canonical routes:
 
 - `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
+- `/evermind`
+- `/nevermind`
+- `/mastermind`
 - `/docs`
 - `/contact`
 - `/privacy`
diff --git a/docs/design/MOTION_STORYBOARD.md b/docs/design/MOTION_STORYBOARD.md
index 80ff8bf..50ca1de 100644
--- a/docs/design/MOTION_STORYBOARD.md
+++ b/docs/design/MOTION_STORYBOARD.md
@@ -3,6 +3,8 @@
 **Status:** canonical motion planning document  
 **Authority:** `docs/homepage-visual-storyboard.md`, `DESIGN.md`
 
+> Historical pre-cinematic storyboard. Its Memory/QA-specific sequences are not the current public route specification. Current routes `/`, `/evermind`, `/nevermind`, and `/mastermind` share two full-viewport sections, an intentional 80vh empty spacer, and one scroll-driven cinematic media layer. Other public routes are outside this visual redesign. See `docs/homepage-validation-plan.md` and `docs/implementation-plan.md` for active behavior and evidence.
+
 ## Purpose
 
 This document defines narrative states before GSAP implementation. Motion code must follow named product states rather than inventing choreography in place.
diff --git a/docs/homepage-validation-plan.md b/docs/homepage-validation-plan.md
index ac5ed55..7e61425 100644
--- a/docs/homepage-validation-plan.md
+++ b/docs/homepage-validation-plan.md
@@ -1,441 +1,45 @@
-# ProChat Homepage Validation Plan
+# ProChat public-site validation plan
 
-**Status:** approved pre-build quality plan  
-**Applies to:** prototypes, production implementation, and release readiness
+**Status:** current validation contract for the cinematic public website
+**Applies visually to:** `/`, `/evermind`, `/nevermind`, and `/mastermind`. Other canonical routes are regression-only.
 
-The homepage is not complete when it looks impressive on one desktop viewport. It is complete when product comprehension, visual quality, motion, accessibility, performance, responsiveness, and claim safety all pass defined checks.
+Product truth comes from `mind/organizations/prochat/brand/`; implementation state comes from current source and `docs/implementation-plan.md`. Older homepage/page-architecture plans are historical unless explicitly reconciled.
 
-## Validation layers
+## Product and content checks
 
-```text
-1. product truth
-2. comprehension
-3. visual quality
-4. responsive design
-5. motion behavior
-6. accessibility
-7. performance
-8. browser compatibility
-9. conversion flow
-10. production safety
-```
+- Evermind is free, open-source, human-owned memory; Nevermind is its paid capability/convenience layer; Mastermind is a separate free planning/orchestration product.
+- ProChat Memory for QA is paused/historical, not a current public product.
+- Home communicates the product family accurately; each product route explains its own role and CTA destination.
+- Product identities on the four cinematic routes match Mind; do not change Contact/Docs copy or navigation in this release.
+- Claims about privacy, security, integrations, pricing, availability, and outcomes match Mind and verified implementation. Privacy/Terms meaning must not be changed by a shell migration.
 
-## 1. Product-truth validation
+## Shared cinematic-page checks
 
-Review every section against:
+For `/`, `/evermind`, `/nevermind`, and `/mastermind` verify:
 
-- `PRODUCT.md`
-- canonical Mind product strategy
-- canonical homepage copy
-- claim boundaries
+- one shared template and one full-width transparent navigation; no legacy capsule, extra header/banner, mixed theme band, or footer;
+- exactly one homepage/page H1 and one main landmark;
+- section one at approximately `100svh`, intentional empty `80vh` spacer, section two at approximately `100svh`, then page end;
+- shared route data changes copy without changing geometry or visual system;
+- capability display is one translucent glass panel, with background media visibly present through it;
+- media visibly scrubs with scroll, forward and reverse; adjacent frames blend and renderer handoffs stay continuously painted;
+- two stages activate in reading order, without text overlap or jarring chapter jumps;
+- every CTA is usable and remains on one line without clipping.
 
-Check that:
+## Responsive and motion matrix
 
-- ProChat has exactly two products;
-- Memory remains the flagship;
-- Memory for QA is an edition, not a third product;
-- Workbench remains the second product;
-- current evidence remains stronger than stored memory;
-- human review precedes trusted memory;
-- local storage claims remain precise;
-- visuals do not imply unsupported integrations, encryption, customer results, or automation.
+Check widths `1440`, `1280`, `1024`, `768`, `430`, `390`, and `360` pixels. Review desktop, laptop, tablet, and mobile screenshots at the top, mid-hero, transition, section two/content, and bottom after CSS transitions settle. Verify text contrast, heading fit, sticky/fixed-nav collisions, CTA/button collisions, capability/readability, section pacing, and layout shifts.
 
-Any visual implication counts as a claim.
+Under `prefers-reduced-motion: reduce`, verify all stages and links remain available, no scroll scrubbing occurs, media freezes to one stable poster/frame, video and canvas are not visibly duplicated, and layout geometry remains understandable. Include a reduced-motion desktop screenshot.
 
-## 2. Comprehension testing
+## Accessibility, performance, and runtime
 
-### Five-second hero test
+- Keyboard, focus-visible, accessible names/landmarks, form labels/errors, and target sizes pass.
+- No horizontal overflow, uncaught page errors, hydration warnings, or missing assets on the four cinematic routes; existing CI guards the other canonical routes.
+- Capture Lighthouse mobile LCP for `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, and `/contact`; target ≤2.5s. Report lab LCP separately from field INP.
+- Evaluate a local LiquidGlass prototype against the current translucent CSS glass for fidelity, initialization cost, continuous frame cost, scroll cadence, accessibility, and reduced motion. Add no dependency unless measured benefit justifies it.
+- Verify deployed `/api/version` equals the intended release SHA and smoke-test all eight routes/known redirects.
 
-Show the static first hero state for five seconds.
+## Release evidence
 
-Ask:
-
-1. What does ProChat do?
-2. What is the main product?
-3. What happens to project knowledge?
-4. Who decides what becomes trusted?
-5. What can someone use today?
-
-Passing summary:
-
-> ProChat keeps reviewed project knowledge reusable in local memory. Its flagship is ProChat Memory, and the current edition is for QA.
-
-Target:
-
-- at least 4 of 5 test participants identify the flagship and core problem;
-- at least 3 of 5 identify human review and current QA availability;
-- no participant describes ProChat as only a chatbot, note-taking app, or autonomous agent.
-
-### Section comprehension
-
-After each cinematic chapter, ask the visitor to explain the state change.
-
-The explanation should match the product mechanism without requiring the accompanying paragraph.
-
-## 3. Visual-quality review
-
-Evaluate each chapter at:
-
-- initial state;
-- middle state;
-- final state;
-- mobile state;
-- reduced-motion state.
-
-Review criteria:
-
-- focal point;
-- hierarchy;
-- typography;
-- line wrapping;
-- spacing rhythm;
-- alignment;
-- product realism;
-- consistency;
-- cobalt restraint;
-- border and depth quality;
-- absence of generic AI patterns;
-- visual continuity;
-- CTA clarity.
-
-A chapter fails if it requires animation to hide a weak static composition.
-
-## 4. Responsive matrix
-
-Required screenshot sizes:
-
-```text
-360×800
-390×844
-768×1024
-1024×768
-1280×800
-1440×900
-1728×1117
-```
-
-Check:
-
-- no horizontal overflow;
-- no unreadable scaled product UI;
-- no clipped headline or CTA;
-- no overlap that conceals content;
-- stable transitions around breakpoints;
-- readable product metadata;
-- sufficient touch targets;
-- purpose-built mobile scenes;
-- no desktop pinning assumptions on mobile.
-
-## 5. Motion validation
-
-### Functional checks
-
-For every cinematic sequence:
-
-- scroll forward through all named states;
-- scroll backward through all named states;
-- jump into the middle through direct navigation or restored scroll;
-- resize while inside the section;
-- rotate a touch device;
-- navigate away and back;
-- verify all ScrollTriggers clean up;
-- verify no pin spacing jump;
-- verify no hidden interactive element receives focus.
-
-### Motion-quality checks
-
-Review:
-
-- motion purpose;
-- timing;
-- easing;
-- continuity;
-- reversibility;
-- perceived weight;
-- reading stability;
-- over-animation;
-- missing transitions;
-- consistency between chapters.
-
-### Reduced motion
-
-With `prefers-reduced-motion: reduce`:
-
-- no pinning;
-- no scrubbed object travel;
-- no parallax;
-- no autoplay loop;
-- every product conclusion remains visible;
-- the page remains visually designed rather than degraded.
-
-## 6. Accessibility validation
-
-Minimum standard:
-
-```text
-WCAG 2.2 AA
-```
-
-Manual checks:
-
-- keyboard-only navigation;
-- visible focus;
-- logical heading order;
-- logical DOM reading order;
-- screen-reader labels for product-state summaries;
-- accordions and dialogs announce state;
-- no state communicated only by color;
-- 200% zoom;
-- high text contrast;
-- meaningful graphical-object contrast;
-- touch targets;
-- no content loss with CSS animation disabled.
-
-Automated checks:
-
-- Playwright tests;
-- `@axe-core/playwright`;
-- HTML validation where available;
-- contrast checks;
-- accessibility tree snapshots for critical sections.
-
-Automated checks supplement manual review and do not replace it.
-
-## 7. Performance validation
-
-Targets:
-
-```yaml
-LCP_seconds: 2.5
-INP_ms: 200
-CLS: 0.1
-hero_media_initial_kb: 250
-above_fold_transfer_kb: 700
-pinned_chapters_max: 4
-total_scroll_triggers_target_max: 24
-```
-
-Measure:
-
-- production build output;
-- JavaScript bundles;
-- GSAP chunk size;
-- font transfer;
-- hero render timing;
-- layout shift during font and animation initialization;
-- long tasks during scroll;
-- dropped frames on a mid-range mobile profile;
-- memory growth after repeated route transitions;
-- interaction latency for navigation, accordions, and CTAs.
-
-Tools:
-
-- Next bundle analyzer;
-- browser Performance panel;
-- Lighthouse;
-- Web Vitals instrumentation;
-- Playwright traces;
-- mobile CPU and network throttling.
-
-## 8. Browser and input matrix
-
-Browsers:
-
-- current Chrome;
-- current Safari;
-- current Firefox;
-- current Edge.
-
-Inputs:
-
-- mouse wheel;
-- trackpad;
-- keyboard;
-- touch;
-- screen reader where available.
-
-Preferences:
-
-- standard motion;
-- reduced motion;
-- light mode;
-- approved dark technical panels.
-
-## 9. Visual regression testing
-
-Use Playwright screenshot assertions.
-
-Create deterministic visual states rather than screenshotting arbitrary points during an active animation.
-
-Each cinematic component should support a development/test state parameter:
-
-```text
-?visualState=current-evidence
-?visualState=human-review
-?visualState=approved-memory
-```
-
-Alternatively, expose internal test helpers only in development and test builds.
-
-Required baselines:
-
-- hero initial, review, and focused-context states;
-- Memory lifecycle evidence, review, approved, and retired states;
-- context-selection workspace and focused states;
-- QA failed-test, investigation, approved-lesson, and later-use states;
-- Workbench request, validation, and committed states;
-- mobile scenes;
-- reduced-motion scenes;
-- FAQ expanded state;
-- focus-visible state;
-- error or blocked status where shown.
-
-## 10. Conversion validation
-
-Validate actions and destinations:
-
-- Explore ProChat Memory for QA
-- Join the selected QA beta
-- See how Memory works
-- Explore ProChat Workbench
-
-Check:
-
-- CTA label predicts the destination;
-- no dead or legacy routes;
-- navigation and footer use current product names;
-- beta form asks only necessary questions;
-- form validation is understandable;
-- success and failure states exist;
-- analytics events are named consistently;
-- no conversion flow implies immediate paid availability.
-
-## 11. Content and metadata validation
-
-Review:
-
-- title;
-- meta description;
-- canonical URL;
-- Open Graph title and image;
-- social description;
-- heading hierarchy;
-- link text;
-- alt text;
-- structured data where relevant;
-- sitemap routes;
-- robots behavior for design-lab routes.
-
-The design-lab route must not appear in public navigation, sitemap, or search indexing.
-
-## 12. Production-safety validation
-
-Before merge:
-
-- no debug ScrollTrigger markers;
-- no design-lab links in production navigation;
-- no prototype query controls exposed publicly;
-- no real customer or client data in examples;
-- no unsupported claims;
-- no accidental public Git references to sensitive memory;
-- no console errors;
-- no hydration warnings;
-- no unbounded animation listeners;
-- no unrelated changes staged;
-- all dependencies have reviewed licenses and versions;
-- production build passes.
-
-## Skill-based review sequence
-
-```text
-/plan-design-review
-→ before production implementation
-
-/design-motion-principles
-→ after each cinematic prototype
-
-/redesign-skill + /code
-→ during integration
-
-/design-review
-→ against the assembled running site
-
-/impeccable audit + polish + harden
-→ final production-quality pass
-```
-
-## Release gates
-
-### Gate A — Foundation
-
-Pass when:
-
-- tokens and typography are approved;
-- color contrast passes;
-- desktop and mobile specimens exist.
-
-### Gate B — Hero
-
-Pass when:
-
-- five-second comprehension passes;
-- static and animated states are approved;
-- reduced motion and mobile are complete;
-- performance proof is acceptable.
-
-### Gate C — Product mechanics
-
-Pass when:
-
-- Memory lifecycle;
-- relevant context;
-- QA investigation;
-- Workbench control plane
-
-all pass product-truth and motion review.
-
-### Gate D — Full-page prototype
-
-Pass when:
-
-- narrative pacing is approved;
-- visual consistency is approved;
-- all routes and CTAs are correct;
-- mobile and reduced motion are coherent.
-
-### Gate E — Production candidate
-
-Pass when:
-
-- build, accessibility, browser, performance, and visual regression tests pass;
-- independent design review findings are resolved;
-- canonical claim review passes.
-
-## Evidence required at completion
-
-```yaml
-release_evidence:
-  screenshots:
-    - desktop
-    - tablet
-    - mobile
-    - reduced_motion
-  tests:
-    - type_check
-    - production_build
-    - playwright_functional
-    - playwright_visual
-    - axe_accessibility
-  performance:
-    - lighthouse
-    - bundle_report
-    - browser_trace
-  reviews:
-    - product_truth
-    - design_review
-    - motion_review
-    - final_polish
-  git:
-    - explicit_changed_paths
-    - isolated_commits
-```
+Retain reproducible test output and manually reviewed viewport screenshots for all four cinematic routes; include desktop, tablet, mobile, reduced-motion, reverse-scroll, frame-cache handoff, and production revision/route evidence. Existing route tests guard unrelated utility/legal pages. Do not mark release complete based only on a local build or push.
diff --git a/docs/implementation-plan.md b/docs/implementation-plan.md
index 21fe7d9..8025ae0 100644
--- a/docs/implementation-plan.md
+++ b/docs/implementation-plan.md
@@ -1,113 +1,48 @@
-# ProChat implementation plan
+# ProChat website implementation plan
 
-Status: lean public implementation complete and deployed; final repository hygiene/hardening pass in progress.
+Status: cinematic/site quality implementation and release validation in progress.
 
-Last consolidated: 2026-08-13.
+Last reviewed: 2026-09-29.
 
-## Completed implementation
+## Current source contract
 
-The deployed public implementation consists of exactly eight canonical routes:
+Canonical public routes: `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, `/contact`, `/privacy`, and `/terms`. Product truth is governed by Mind at `mind/organizations/prochat/brand/`; the current identities are Evermind, Nevermind, and Mastermind. ProChat Memory for QA is paused/historical.
 
-- `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
-- `/docs`
-- `/contact`
-- `/privacy`
-- `/terms`
+The first four routes use a shared cinematic implementation: `HomeV2` or `CinematicProductFunnel` supplies page data to `CinematicMarketingPage`; the shared page uses `CinematicMarketingShell`, one final cinematic reference stylesheet, and `ScrollVideoBackground`. Route content must not reintroduce the former public-site shell, extra headers, a footer, opaque SaaS cards, or competing route-level overrides.
 
-The completed lean migration removed retired product/catalog/commerce/content systems and consolidated the active design/runtime boundaries.
+Contact, Docs, Privacy, and Terms retain their existing route-specific presentation and are outside this cinematic redesign. Keep their established implementation and legal meaning unchanged; use the repository's existing route checks to guard against regressions.
 
-Major completed implementation areas:
+## Active work gates
 
-- canonical route shell and route-specific body evidence;
-- Memory, Memory for QA, and Workbench product pages;
-- lean Docs repository hub;
-- Contact and beta-interest handling;
-- current legal pages;
-- root sitemap and robots policy;
-- centralized graphite/cobalt design tokens across public and active admin surfaces;
-- removal of generated Docs/Nextra, dynamic legacy Blog content, Prompts/Learn/Production Guide, Kits, ProChat OS, AI Workflows, Studio, Proof, checkout, Stripe runtime, application licensing, MailerLite, GitHub entitlement provisioning, and other verified zero-consumer legacy code;
-- compatibility redirects from historical URLs into current canonical routes;
-- security/browser/accessibility evidence for current public routes and retired-route behavior;
-- CI/Dokploy deployment and exact production SHA verification.
+### A. Renderer and motion
 
-## Current final hygiene implementation
+- Reproduce slow/fast forward and reverse scroll in a real H.264-capable browser.
+- Prove cache and bootstrap handoffs do not expose blank frames; compare frames through blending and stop painting when settled.
+- Confirm the requested background demonstrably changes with scroll and chapter activation follows reading order.
+- Under reduced motion, freeze to one stable media/poster layer, disable scrubbing, preserve all content/links, and retain geometry.
 
-The post-release hygiene pass is deliberately non-product-changing. It may update repository code/docs only where current-source evidence proves cleanup or documentation drift.
+### B. Visual and content QA
 
-### H1 — orphan code and local artifacts
+- Check all four cinematic routes at desktop, laptop, tablet, and mobile widths; verify one H1, one main, no footer, no horizontal overflow, CTA wrapping, and consistent 100svh / 80vh / 100svh structure.
+- Do not redesign Contact, Docs, Privacy, or Terms in this release. Preserve their current route and legal behavior through existing regression tests.
+- Review current copy against canonical Mind source and prevent stale product identities/claims in active pages and docs.
+- Capture and manually inspect viewport screenshots after motion settles, including top, transition, Section Two, page end, and reduced-motion desktop.
+- Prototype LiquidGlass locally and compare quality, CPU/frame cadence, media scrubbing, reduced motion, and initialization cost. Do not add a dependency or ship the effect without measured benefit.
 
-- delete zero-consumer legacy helpers;
-- keep dependencies that still have active consumers;
-- remove ignored Playwright reports/test-result artifacts where available tooling permits;
-- do not alter canonical public behavior.
+Initial LiquidGlass evaluation: a real Chrome local prototype using one dynamic scroll-canvas background and one cloned context panel ran at approximately 11–15fps (median frame interval 50–66.6ms) compared with the existing renderer's 16.7ms median cadence. The captured overlay also clipped/reflowed the cloned panel. Decision: do not add the dependency to the production scroll path; current CSS glass uses an 11% translucent white surface with 18px backdrop blur and preserves media throughput. This is a measured rejection for the current implementation, not a claim that the library cannot be used on static pages.
 
-### H2 — documentation alignment
+### C. Validation and release
 
-Active documentation must match current source and must not instruct contributors to use retired Stripe, MailerLite, GitHub entitlement, Kits, generated Docs, Strapi, Make/n8n integration credentials, or old product flows.
+Run the repository-supported env/doc integrity, TypeScript, ESLint, design lint, asset validation, production build, security, and relevant browser evidence. Review a diff-based safety pass; stage exact files only; commit and push the continuation branch without rewriting history. Integrate through the repository's normal `main` workflow. Verify workflow and deployment success, `/api/version` full SHA, all eight public routes, media and browser health. Then verify the authoritative worktree is clean and the protected checkout's branch, HEAD, status, and diff fingerprint are unchanged.
 
-Historical implementation documents may remain under `docs/archive/**` and `docs/migration/**`, clearly treated as history rather than current guidance.
+## Known implementation/build debt
 
-### H3 — auth boundary
+`Dockerfile` currently sets placeholder `STRIPE_SECRET_KEY_LIVE` and `STRIPE_WEBHOOK_SECRET_LIVE` values as builder-stage environment variables so retired Stripe module evaluation can pass. They are not real secrets, but they trigger build security warnings. Track a separate narrowly tested refactor to remove this build-time configuration dependency. Never place live credentials in Docker build args or image layers.
 
-Decision: **defer runtime Ory session authorization for internal routes**.
+## Next performance gate
 
-The following remain fail-closed pending a separate security implementation:
+Measure Lighthouse mobile LCP on `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, and `/contact`. The target remains ≤2.5 seconds. If missed, the next roadmap item is to bring canonical public-route LCP under 2.5 seconds without reducing the cinematic experience.
 
-- `/admin/**`
-- project APIs
-- Make APIs
-- n8n APIs
-- related internal scenario/active/link routes covered by security tests
+## Completion evidence
 
-No hygiene change may replace 501/misconfigured behavior with unauthenticated access.
-
-### H4 — branches
-
-Non-main branches may be deleted only when merge evidence is available and the branch is not intentionally retained. Do not guess branch ancestry or force-delete.
-
-## Validation gate
-
-Before this hygiene pass may be committed/pushed:
-
-```bash
-node scripts/check-env-docs.js
-node scripts/check-doc-links.js
-npm run typecheck
-npm run lint
-npm run lint:design
-npm run build
-```
-
-Also run:
-
-- security API tests;
-- focused canonical browser smoke/evidence sufficient to prove public behavior did not regress;
-- final source/reference scan for stale active integrations/products;
-- high-risk and secret-material scans on changed active files;
-- final diff/status review.
-
-## Commit/push gate
-
-Commit only after all required validation passes. Stage exact paths only. Do not include ignored browser artifacts or unrelated work.
-
-After push/deployment where applicable:
-
-- verify production `/api/version` equals the intended full commit SHA;
-- verify the eight canonical routes;
-- verify key compatibility redirects and retired-route behavior;
-- require a clean local worktree.
-
-## Definition of done
-
-This repository can be called fully closed out when:
-
-1. active code/docs contain no known stale retired-runtime guidance;
-2. zero-consumer cleanup is complete for the audited scope;
-3. current status/roadmap/implementation docs agree;
-4. deferred Ory functionality is explicitly documented and remains fail-closed;
-5. branch cleanup is either completed with evidence or reported as the only external/manual limitation;
-6. validation passes;
-7. the hygiene commit is pushed and the worktree is clean.
+Record test commands/results, screenshot review, measured LiquidGlass comparison, measured LCP, feature and production commits, workflow/deployment result, live version and route checks, final branch/worktree state, preserved-checkout fingerprint, and any remaining warnings. Do not call the work complete on a green local build or successful push alone.
diff --git a/docs/overview.md b/docs/overview.md
index 019433a..00f7125 100644
--- a/docs/overview.md
+++ b/docs/overview.md
@@ -20,23 +20,23 @@ Start here for present-day ProChat behavior:
 - `docs/ai-guidelines.md` — current AI/automation repository safety rules
 - `docs/content-platform.md` — current public content/SEO contract
 - `docs/docs-automation.md` — explicit retirement record for the generated Docs system
-- `docs/product/agent-mode-progress.md` — latest operational handoff/current state
+- `docs/product/agent-mode-progress.md` — historical 2026-08-14 closeout, superseded by `docs/repo-status.md`
 - `docs-public/environment.md` — active environment-variable contract
 
 ## Canonical public website
 
-The public site is intentionally limited to:
+The current public site surface is:
 
 - `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
+- `/evermind`
+- `/nevermind`
+- `/mastermind`
 - `/docs`
 - `/contact`
 - `/privacy`
 - `/terms`
 
-Memory, Memory for QA, and Workbench are the current product identities.
+Evermind, Nevermind, and Mastermind are the current public product identities. ProChat Memory for QA is paused/historical. Home and the three product pages share the current cinematic navigation and scroll-driven media. Docs, Contact, and legal pages retain their existing page-specific presentation; they are outside this cinematic redesign scope.
 
 ## Historical evidence
 
@@ -58,7 +58,7 @@ The final post-release hygiene pass additionally archives several retired system
 
 ## Compatibility wording
 
-BuildFlow may remain only as a technical/internal compatibility identifier required by Workbench contracts. Historical redirects may still mention retired product names in tests/config so inbound links resolve safely.
+BuildFlow may remain only as a technical/internal compatibility identifier where required. Historical redirects/tests may mention retired product names so inbound links resolve safely.
 
 ## Documentation rule
 
diff --git a/docs/platform/PERFORMANCE_STRATEGY.md b/docs/platform/PERFORMANCE_STRATEGY.md
index fe24cac..45ef1fd 100644
--- a/docs/platform/PERFORMANCE_STRATEGY.md
+++ b/docs/platform/PERFORMANCE_STRATEGY.md
@@ -2,6 +2,8 @@
 
 **Status:** canonical repository-local performance program
 
+> Current-site note (2026-09-29): the homepage and three product routes now use a scroll-driven video-to-canvas renderer with a stable static bootstrap and reduced-motion fallback. This approved exception to the old static-frame-only plan must be judged against scroll smoothness, reverse-scroll continuity, cache handoff, mobile LCP, and reduced motion. PXF-016D/016D2 measurements below describe the earlier route/theme architecture; rerun them for the current eight public routes before treating their results as current.
+
 ## Principle
 
 Performance is a product and design requirement. Budgets constrain fonts, media, JavaScript, animation, and component architecture before implementation.
@@ -60,7 +62,7 @@ The following items cannot be verified by Lighthouse and require field data or m
 - Lazy-load below-the-fold cinematic modules.
 - Native scrolling only.
 - Semantic HTML, CSS, and SVG for essential visuals.
-- No autoplay hero video or frame sequence.
+- No autoplay hero video. A user-scroll-driven frame sequence is permitted only with continuous fallback, no blank cache handoff, mobile performance evidence, and a stable reduced-motion mode.
 - No smooth-scroll dependency by default.
 
 ## Fonts
@@ -101,7 +103,7 @@ Before adding a dependency, document:
 - maintenance and licence;
 - removal plan.
 
-Planned additions such as GSAP, Playwright, and axe occur in their approved implementation batch, not during documentation or static prototype work.
+Do not add a production dependency for a static prototype. Playwright is already used for browser evidence; any new renderer/glass dependency requires the cost, accessibility, maintenance, and removal review above.
 
 ## Measurement
 
@@ -146,7 +148,7 @@ selection_reason: >
 
 ## Deferred security observation (PXF-016D)
 
-CI run 30758962840 reported two Dockerfile build annotations concerning Stripe live-secret names in Docker `ARG`/`ENV` directives. These are classified as deferred container and secret-handling hardening work. They do not represent plaintext secrets committed to the repository. No action is required in PXF-016D.
+CI run 30758962840 reported Dockerfile annotations concerning Stripe live-secret names in Docker build directives. Current source still declares placeholder `STRIPE_SECRET_KEY_LIVE` and `STRIPE_WEBHOOK_SECRET_LIVE` values using builder-stage `ENV` so retired Stripe module evaluation succeeds at build time. These are placeholders, not real credentials, but their presence implies an obsolete build contract. Track removal of this build-time configuration dependency; never pass actual secrets into Docker build args or layers.
 
 
 ## LCP attribution semantics (PXF-016D2)
@@ -163,14 +165,17 @@ The canonical runner therefore records both:
 
 Do not claim a hydration delay merely because simulated LCP is later than simulated FCP. PXF-016D2 measured the failing public text candidates at or within approximately 17ms of raw FCP, rejecting the earlier 1.7–2.0 second hydration-floor hypothesis.
 
-The current blocking result is six canonical routes above the unchanged 2.5-second simulated LCP threshold. Two bounded font-loading experiments were completed and reverted because they produced no consistent material improvement.
+The historical blocking result was six then-canonical routes above the unchanged 2.5-second simulated LCP threshold. It is not a measurement of the current cinematic routes. Rerun the diagnostic and report each currently canonical route separately before selecting an optimization.
+
+### LiquidGlass evaluation (2026-09-29)
+
+A local system-Chrome prototype placed one LiquidGlass element over a scroll-driven background canvas and marked that canvas changed on each animation frame, as required for live canvas pixel updates. At 1440×900, two repeated scroll runs measured the effect at roughly 11–15fps with 50–66.6ms median RAF intervals; the existing renderer's median remained 16.7ms. The prototype also failed to preserve the context panel's original text composition. It is not approved for the cinematic scroll path. Existing CSS glass remains the current direction unless a future measured test demonstrates a faster and faithful integration.
 
 The next approved repair boundary is architectural critical-path reduction:
 
 1. separate canonical public CSS from legacy and protected-route CSS now carried by the shared render-blocking bundle;
 2. preserve only measured above-the-fold canonical styles in the initial route path;
-3. address the `/docs` Nextra/HeadlessUI transfer and two-candidate LCP separately;
+3. evaluate `/docs` transfer and LCP from current route-source evidence; generated Docs/Nextra is no longer the public Docs route;
 4. avoid root provider or shell rewrites unless later trace evidence proves they delay observed paint.
 
-Phase 12 remains partial until all eight canonical routes pass the unchanged Linux CI laboratory thresholds. Field INP and manual accessibility remain separate evidence requirements.
-
+Historical Phase 12 status does not establish current cinematic release readiness. Field INP and manual accessibility remain separate evidence requirements.
diff --git a/docs/product/PUBLIC_PAGE_ARCHITECTURE.md b/docs/product/PUBLIC_PAGE_ARCHITECTURE.md
index d1218e7..c0e8419 100644
--- a/docs/product/PUBLIC_PAGE_ARCHITECTURE.md
+++ b/docs/product/PUBLIC_PAGE_ARCHITECTURE.md
@@ -1,4 +1,6 @@
-# ProChat Public Page Architecture
+# ProChat Public Page Architecture — superseded historical plan
+
+> Historical plan dated 2026-07-19. Its Memory/Memory-for-QA/Workbench hierarchy, `/memory` routes, page architecture, and acceptance criteria are no longer current. For current route/product guidance see `PRODUCT.md`, `docs/overview.md`, `docs/roadmap.md`, and `docs/implementation-plan.md`. Mind at `mind/organizations/prochat/brand/` remains canonical for product truth.
 
 **Status:** CANONICAL PUBLIC-EXPERIENCE ROUTE AND PAGE-PURPOSE CONTRACT  
 **Decision date:** 2026-07-19  
diff --git a/docs/product/agent-mode-progress.md b/docs/product/agent-mode-progress.md
index fa98f55..3d03df6 100644
--- a/docs/product/agent-mode-progress.md
+++ b/docs/product/agent-mode-progress.md
@@ -1,4 +1,6 @@
-# ProChat agent-mode progress
+# ProChat agent-mode progress — superseded historical closeout
+
+> Historical closeout written 2026-08-14. Its `/memory`, `/memory-qa`, `/workbench` routes, product names, and “hygiene complete” status predate the 2026-09-11 product decision and current cinematic website. Do not use this file as current operational state. Current guidance: `docs/repo-status.md`, `docs/roadmap.md`, and `docs/implementation-plan.md`.
 
 **Current status:** lean public release deployed and production-verified; repository-scoped post-release hygiene/hardening is complete and validated.
 
diff --git a/docs/repo-status.md b/docs/repo-status.md
index bbac810..b6aa4d7 100644
--- a/docs/repo-status.md
+++ b/docs/repo-status.md
@@ -1,79 +1,30 @@
 # Repository status
 
-Status: lean public release deployed; final repository hygiene/hardening pass in progress.
+Status: current source has a locally validated cinematic implementation; production revision and route health are verified through the normal `main` workflow and `/api/version`.
 
-Last consolidated: 2026-08-13.
+Last reviewed: 2026-09-29.
 
-## Public product surface
+## Canonical public surface
 
-Current canonical routes:
+Routes: `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, `/contact`, `/privacy`, and `/terms`.
 
-- `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
-- `/docs`
-- `/contact`
-- `/privacy`
-- `/terms`
+Evermind and Nevermind are the connected Infinite Brain product system; Mastermind is the separate free planning/orchestration product. ProChat Memory for QA is paused/historical. Mind is canonical for product naming, hierarchy, claims, and company strategy.
 
-Memory, Memory for QA, and Workbench are the active product identities. BuildFlow remains only where a technical/internal compatibility identifier is required by Workbench contracts.
+## Runtime and visual architecture
 
-## Retired runtime surfaces
+- `/`, `/evermind`, `/nevermind`, `/mastermind`: shared cinematic template, scroll-driven media, two chapters, 80vh intentional spacer, one navigation, and no footer.
+- `/contact`, `/docs`, `/privacy`, `/terms`: existing route-specific presentation, outside this visual redesign and checked for regressions only.
+- Production integration: GitHub Actions `.github/workflows/main.yml` on `main`; production revision is checked through `/api/version`.
 
-The following are not active public products or runtime systems:
+## Historical or deferred systems
 
-- ProChat OS
-- SaaSKit, ProKit, UXKit, WaaSKit
-- AI Workflows / Legal AI Workflows
-- Studio, Proof, Prompts, Learn, Production Guide
-- generated Docs/Nextra and dynamic legacy Blog content
-- Stripe checkout/webhook/subscription/portal runtime
-- application licence administration/provisioning
-- MailerLite funnel integration
-- GitHub purchaser entitlement provisioning
-- WordPress/FluentCRM inside this Next.js repository
+Generated Docs/Nextra, retired checkout/licensing, Stripe runtime, MailerLite, GitHub purchaser provisioning, and legacy product identities are not current public-site capabilities. Ory runtime authorization for internal/admin surfaces remains separately deferred and fail-closed. Historical references live under archives/migration docs and Git history.
 
-Historical evidence remains under `docs/archive/**`, `docs/migration/**`, and Git history.
+## Validated release scope
 
-## Active runtime capabilities
+Renderer reverse-scroll/cache-handoff, responsive/reduced-motion and interaction evidence, measured LiquidGlass evaluation, current copy/documentation reconciliation, production build, browser/security checks, and manual desktop/mobile review are recorded for the cinematic release. Integrate through the normal GitHub `main` workflow and verify the deployed revision/routes before describing a publication as live. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
 
-- canonical public marketing/product/legal pages
-- Contact API
-- beta-interest/waitlist API and compatibility alias
-- Umami analytics
-- Resend-backed Contact/beta-interest email when configured
-- tenant/database scripts and Prisma runtime
-- shared sign-in/sign-up UI using Ory browser flows
-- internal social automation protected by `SOCIAL_AUTOMATION_SECRET`
+## Tracked performance/security notes
 
-## Deferred internal functionality
-
-Runtime Ory session validation is not implemented for `/admin`, project, Make, and n8n APIs. Those capabilities intentionally fail closed with 501/misconfigured responses.
-
-This is an explicit deferred-hardening decision and does not block the canonical public website.
-
-## Environment contract
-
-The active environment contract is defined by:
-
-- `.env.example`
-- `docs-public/environment.md`
-
-Do not reintroduce retired Stripe/GitHub/MailerLite/Strapi/Make/n8n/generated-Docs credentials without a separately approved runtime consumer.
-
-## Validation baseline
-
-The lean release has previously passed production build, TypeScript, ESLint, design lint, security tests, canonical browser/accessibility evidence, deployment SHA verification, desktop/mobile production route checks, compatibility redirects, and retired-route checks.
-
-The final hygiene batch must rerun the relevant validation before commit/push because source/docs changed again.
-
-## Current hygiene scope
-
-- remove verified orphan helpers;
-- archive retired-system docs that were still in active locations;
-- align active documentation with code;
-- explicitly document the deferred Ory boundary;
-- remove ignored browser artifacts where possible;
-- verify branch cleanup with evidence before deletion;
-- validate, commit, push, and require a clean worktree.
+- Canonical mobile LCP target: ≤2.5s; capture actual route measurements before setting the next goal.
+- Docker builder still declares placeholder `STRIPE_SECRET_KEY_LIVE` and `STRIPE_WEBHOOK_SECRET_LIVE` values via `ENV`; these are not credentials, but trigger container-build security warnings. Remove the obsolete build-time dependency in a separately validated change, without passing real secrets to the builder.
diff --git a/docs/roadmap.md b/docs/roadmap.md
index e0162e1..97389e6 100644
--- a/docs/roadmap.md
+++ b/docs/roadmap.md
@@ -1,81 +1,45 @@
-# ProChat roadmap
+# ProChat website roadmap
 
-Status: lean public-site roadmap complete; only explicitly deferred internal hardening remains.
+Status: cinematic implementation locally validated; use the normal `main` workflow and `/api/version` to determine the published revision. Performance and dependency hardening remain separate follow-up goals.
 
-Last consolidated: 2026-08-13.
+Last reviewed: 2026-09-29.
 
-## Completed public roadmap
+## Current public surface
 
-The public ProChat site has been consolidated to exactly eight canonical routes:
+The active public routes are `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, `/contact`, `/privacy`, and `/terms`. Evermind and Nevermind are the connected Infinite Brain product system; Mastermind is the separate free planning/orchestration product. ProChat Memory for QA is paused/historical. Mind remains canonical for product and company strategy.
 
-- `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
-- `/docs`
-- `/contact`
-- `/privacy`
-- `/terms`
+The first four routes share a two-section cinematic page, one full-width navigation, one scroll-driven background renderer, one 80vh spacer, and no footer. Contact, Docs, Privacy, and Terms are regression-checked but not redesigned in this release; preserve their current route presentation and legal meaning.
 
-The production release is deployed from `main` and was verified at the exact deployed Git SHA after the lean cleanup.
+## Current release goal — ProChat cinematic/site quality pass
 
-Completed outcomes include:
+The release source of truth is `feature/prochat-cinematic-finalization-2026-09`; local validation was performed in the managed `prochat-quality-final` worktree. Do not modify the protected dirty checkout at `/Users/Office/Repos/prochattools/web/prochat`.
 
-- Memory, Memory for QA, and Workbench as the active product surface;
-- lean Docs hub for Memory for QA and Workbench repositories;
-- Contact and beta-interest flow;
-- current Privacy and Terms pages;
-- canonical sitemap/robots/SEO cleanup;
-- one graphite/cobalt design-token system;
-- retired Kits, ProChat OS, AI Workflows, Studio, Proof, Prompts, generated Docs/Nextra, dynamic Blog content, checkout, Stripe runtime, application licensing, MailerLite, and GitHub entitlement provisioning;
-- compatibility redirects into current canonical routes;
-- production desktop/mobile route verification, security tests, accessibility/browser evidence, build/type/lint/design checks, and deployment SHA verification.
+In scope:
 
-## Repository hygiene roadmap
+1. Fix and verify smooth reverse/forward scroll, cache/bootstrap frame blending, chapter pacing, and CTA wrapping on the four cinematic product routes.
+2. Keep those four routes visually consistent, including Golos display headings, one navigation, translucent capability glass, correct 100svh / 80vh / 100svh geometry, and zero footers.
+3. Keep Contact, Docs, Privacy, and Terms outside the visual redesign; retain existing routes and verify their established CI contracts.
+4. Verify prefers-reduced-motion, 1440/1280/tablet/mobile layouts, route content, menu interactions, browser errors, and manually inspected cinematic screenshots.
+5. Evaluate LiquidGlass with a measured local prototype; do not ship it unless it materially improves the glass treatment without harming media scroll performance, accessibility, or initial rendering. Initial 1440×900 Chrome prototype result: not accepted for this scroll-driven page—the one-panel dynamic-canvas run measured about 11–15fps (50–66.6ms median RAF interval) versus 16.7ms baseline cadence, and the prototype overlay did not preserve the card composition. Keep the current 11% CSS liquid-glass treatment.
+6. Audit current public copy against Mind, correct active website guidance, validate, review, commit, push, merge through the normal `main` workflow, then verify the live revision and routes.
+7. Preserve the protected checkout and unrelated repositories/worktrees. After confirmed production success only, reconcile Mastermind source continuity, indexes, and safely removable branches/worktrees; do not delete unique or unverified work.
 
-The final post-release hygiene pass is intentionally bounded to repository cleanup and documentation alignment. It must not change canonical public behavior without a concrete defect.
+## Performance follow-up
 
-Current hygiene work:
+The canonical mobile LCP target is at most 2.5 seconds. A single-run local production-artifact diagnostic on 2026-09-29 (Lighthouse 12.4, Moto G Power 2022 simulation, 4× CPU, slow 4G; diagnostic mode) measured LCP as follows: `/` 2.83s, `/evermind` 3.19s, `/nevermind` 2.79s, `/mastermind` 2.80s, `/docs` 2.72s, `/contact` 2.99s. All six exceeded target; CLS was 0 and TBT was 0ms. These are local lab measurements, not production field data, and one run per route is directional rather than a stable baseline. The next website goal is: **Bring canonical public-route LCP under 2.5 seconds without reducing cinematic fidelity.** Do not trade away the requested scroll-driven visuals to satisfy the metric.
 
-1. remove verified zero-consumer helpers and stale environment guidance;
-2. archive retired-system documentation that could be mistaken for active guidance;
-3. align active environment, onboarding, development, analytics, deployment, roadmap, and implementation docs with current code;
-4. preserve archive/migration history as historical evidence;
-5. remove ignored local browser-test artifacts where tooling permits;
-6. verify and clean obsolete branches only after merge evidence;
-7. rerun full validation before committing/pushing the hygiene batch.
+## Tracked build hardening
 
-## Explicitly deferred internal hardening
+The current `Dockerfile` builder stage declares placeholder values for `STRIPE_SECRET_KEY_LIVE` and `STRIPE_WEBHOOK_SECRET_LIVE` with `ENV` so build-time module evaluation succeeds. They are placeholders, not credentials, but the names trigger container-build security annotations and imply an obsolete Stripe build contract. Remove that build-time configuration dependency in a separate verified code change; never pass real Stripe secrets through Docker build arguments or layers. Keep this warning visible until the warning is eliminated and a clean production build is verified.
 
-Runtime Ory session validation for internal `/admin`, project, Make, and n8n routes is not part of the lean public release.
+The 2026-09-29 `npm audit --omit=dev` scan also reported 68 findings (4 critical, 32 high, 30 moderate, 2 low). Next.js 14.2.31 is within the advisory range for AVIF-optimizer RCE, while the current `next.config.js` does not opt into AVIF output; the separately reported Windows-filesystem RCE does not match the Debian-based container runtime. This visual release does not upgrade dependencies; track a scoped dependency-security review and upgrade to supported patched versions before enabling AVIF or changing the runtime platform.
 
-Those routes remain fail-closed with HTTP 501 or equivalent misconfigured behavior. A future roadmap item may implement authenticated Ory session retrieval and authorization, but only with a dedicated security design and tests.
+## Deferred, separate work
 
-This deferred work does **not** block the canonical public website.
+- Runtime Ory authorization for internal/admin APIs remains explicitly fail-closed and requires its own security design.
+- Nevermind product planning, billing/commercialization, and Mind/Brain changes are outside this website release.
+- Do not redesign unrelated product or legal content while performing the cinematic shell migration.
 
-## Closed / not on roadmap
+## Release gate
 
-The following are not active roadmap items unless explicitly re-approved:
-
-- public Kits or paid-kit commerce;
-- Stripe checkout/webhook/subscription runtime;
-- application licence management or entitlement provisioning;
-- MailerLite funnel integration;
-- GitHub App purchaser provisioning;
-- generated public Docs/Nextra pipeline;
-- public ProChat OS, AI Workflows, Studio, Proof, or Prompt products;
-- BuildFlow as a public product identity;
-- WordPress/FluentCRM inside this Next.js runtime.
-
-Historical plans remain in `docs/archive/**` and `docs/migration/**` for evidence only.
-
-## Release gate for future roadmap work
-
-Any new roadmap item that changes public behavior must define:
-
-- current-source evidence;
-- exact route/API/data scope;
-- security/privacy impact;
-- migration/rollback behavior where needed;
-- TypeScript, ESLint, design-lint, build, and relevant security/browser evidence;
-- production SHA and route verification after deployment.
+Completion requires current-source evidence, clean TypeScript/ESLint/design/asset checks, the production build, relevant security and browser suites, manual desktop/tablet/mobile/reduced-motion review, exact reviewed-file staging, normal GitHub integration and deployment, production `/api/version` equality, route checks, and a clean authoritative implementation worktree. A successful local build or push alone is not release completion.
diff --git a/docs/strategy.md b/docs/strategy.md
index a8b3171..ecda95d 100644
--- a/docs/strategy.md
+++ b/docs/strategy.md
@@ -11,40 +11,39 @@ Mind is canonical for ProChat philosophy, product hierarchy, naming, positioning
 Canonical references:
 
 ```text
-mind/wiki/organisations/prochat/brand/README.md
-mind/wiki/organisations/prochat/brand/product-strategy.md
-mind/wiki/organisations/prochat/brand/product-naming-architecture.md
-mind/wiki/organisations/prochat/brand/product-roadmap.md
-mind/wiki/organisations/prochat/brand/canonical-homepage-copy.md
-mind/wiki/organisations/prochat/brand/global-design-foundation.md
-mind/wiki/organisations/prochat/brand/website-visual-motion-system.md
-mind/wiki/organisations/prochat/brand/website-build-contract.md
+mind/organizations/prochat/brand/README.md
+mind/organizations/prochat/brand/product-strategy.md
+mind/organizations/prochat/brand/product-naming-architecture.md
+mind/organizations/prochat/brand/product-roadmap.md
+mind/organizations/prochat/brand/canonical-homepage-copy.md
+mind/organizations/prochat/brand/global-design-foundation.md
+mind/organizations/prochat/brand/website-visual-motion-system.md
+mind/organizations/prochat/brand/website-build-contract.md
 ```
 
 This document is not a second ProChat strategy source. It only explains how this website repository should apply Mind's strategy in local pages, navigation, metadata, and conversion flows.
 
 ## Current product boundary for the website
 
-Mind currently defines exactly two ProChat products:
+Mind's current public-product system is:
 
 ```text
-ProChat Memory
-ProChat Workbench
+Infinite Brain: Evermind ↔ Nevermind
+Mastermind (separate ProChat product)
 ```
 
-The current launch focus is:
+ProChat Memory for QA is paused/historical, not a current public product. The current product routes are:
 
 ```text
-ProChat Memory for QA
+/evermind · /nevermind · /mastermind
 ```
 
 Website priorities:
 
-1. Lead with ProChat Memory as the flagship product.
-2. Present ProChat Memory for QA as the first launch niche and first discipline-specific edition.
-3. Present ProChat Workbench as the second product, separate from Memory.
-4. Keep future capabilities and interfaces clearly labeled as non-products.
-5. Archive or de-emphasize older directions instead of deleting historical material.
+1. Explain Evermind as human-owned memory and Nevermind as the paid capability/convenience layer around it.
+2. Present Mastermind as a separate planning/orchestration product.
+3. Keep future capabilities and interfaces clearly labeled as non-products.
+4. Archive or de-emphasize older directions instead of deleting historical material.
 
 ## What this repository may decide
 
@@ -78,10 +77,10 @@ Those decisions live in Mind.
 
 Use Mind for the canonical product story, then translate it into website surfaces as follows:
 
-- homepage: explain ProChat through the Memory-first product direction without duplicating Mind's full strategy;
-- ProChat Memory page: explain the flagship product and local, inspectable, review-first memory benefits;
-- ProChat Memory for QA page: explain the first launch niche for QA testers;
-- ProChat Workbench page: explain the local ChatGPT-first builder workbench;
+- homepage: explain Evermind, Nevermind, and Mastermind as a connected product family without duplicating Mind's full strategy;
+- `/evermind`: explain free, open-source, human-owned memory;
+- `/nevermind`: explain paid capability and convenience around human-owned memory;
+- `/mastermind`: explain the separate planning/orchestration product;
 - contact/waitlist: route users toward the relevant product or selected beta interest;
 - docs/learn: support understanding and adoption without inventing new product categories.
 
@@ -93,7 +92,7 @@ API access and MCP integrations may be described only as future interfaces or in
 
 ProChat OS, SaaSKit, ProKit, UXKit, WaaSKit, and MikeOSS are legacy, historical, external, or archived references where relevant. They are not current ProChat products.
 
-BuildFlow may remain only as a technical/internal compatibility identifier for Workbench where required.
+BuildFlow may remain only as a technical/internal compatibility identifier where required.
 
 ## Copy safety rules
 
diff --git a/src/components/CinematicMarketingPage.tsx b/src/components/CinematicMarketingPage.tsx
index 3edec39..3d80919 100644
--- a/src/components/CinematicMarketingPage.tsx
+++ b/src/components/CinematicMarketingPage.tsx
@@ -8,7 +8,6 @@ import {
 } from './CinematicMarketingShell'
 import { CinematicChapter, CinematicGlassPanel } from './cinematic/CinematicChapter'
 import { CinematicMediaPage } from './cinematic/CinematicMediaPage'
-import './cinematic/cinematic-reference-final.css'
 
 export type CinematicMarketingCapability = {
   title: string
diff --git a/src/components/CinematicMarketingShell.tsx b/src/components/CinematicMarketingShell.tsx
index fc480de..8c985b5 100644
--- a/src/components/CinematicMarketingShell.tsx
+++ b/src/components/CinematicMarketingShell.tsx
@@ -7,8 +7,8 @@ import { useEffect, useRef, useState } from 'react'
 import Logo from '@/components/logo'
 import { SocialIcon } from '@/components/ui/social-icons'
 import { trackEvent } from '@/utils/analytics'
-
 import './cinematic-marketing-shell.css'
+import './cinematic/cinematic-reference-final.css'
 
 const NAV_ITEMS = [
   { href: '/evermind', label: 'Evermind' },
diff --git a/src/components/cinematic/cinematic-marketing.css b/src/components/cinematic/cinematic-marketing.css
index cc03b17..beb111f 100644
--- a/src/components/cinematic/cinematic-marketing.css
+++ b/src/components/cinematic/cinematic-marketing.css
@@ -38,7 +38,7 @@
 .cm-lede { max-width: 38rem; margin: 1.5rem 0 0; color: var(--cm-muted); font-size: clamp(1rem, 1.6vw, 1.35rem); line-height: 1.65; }
 .cm-chapter--center .cm-lede { margin-inline: auto; }
 .cm-actions { display: flex; flex-wrap: wrap; gap: 0.75rem; margin-top: 2rem; }
-.cm-actions a { min-height: 2.9rem; display: inline-flex; align-items: center; gap: 0.5rem; padding: 0.75rem 1.1rem; border-radius: 999px; font-size: 0.82rem; font-weight: 650; text-decoration: none; }
+.cm-actions a { min-height: 2.9rem; display: inline-flex; flex: 0 0 auto; align-items: center; gap: 0.5rem; padding: 0.75rem 1.1rem; border-radius: 999px; font-size: 0.82rem; font-weight: 650; text-decoration: none; white-space: nowrap; }
 .cm-actions__primary { background: rgb(var(--pc-text-rgb)); color: rgb(var(--pc-bg-rgb)); }
 .cm-actions__secondary { border: 1px solid var(--cm-line); background: rgb(255 255 255 / 0.08); color: rgb(var(--pc-text-rgb)); backdrop-filter: blur(14px); }
 .cm-glass-panel { border: 1px solid var(--cm-line); border-radius: 1.35rem; background: rgb(8 12 19 / 0.48); box-shadow: inset 0 1px 0 rgb(255 255 255 / 0.09), 0 1.5rem 4rem rgb(0 0 0 / 0.22); backdrop-filter: blur(18px); }
@@ -56,7 +56,7 @@
 .cm-choice-grid { grid-template-columns: repeat(3, 1fr); margin-top: 3rem; text-align: left; }
 .cm-choice-grid h3 { margin: 0 0 0.7rem; font-size: 1.3rem; }
 .cm-choice-grid a { display: inline-flex; margin-top: 1.5rem; color: rgb(var(--pc-text-rgb)); font-size: 0.82rem; font-weight: 650; text-decoration: none; }
-.cm-inline-link { display: inline-flex; align-items: center; gap: .45rem; margin-top: 1.5rem; color: rgb(var(--pc-text-rgb)); font-size: .82rem; font-weight: 650; text-decoration: none; }
+.cm-inline-link { display: inline-flex; flex: 0 0 auto; align-items: center; gap: .45rem; margin-top: 1.5rem; color: rgb(var(--pc-text-rgb)); font-size: .82rem; font-weight: 650; text-decoration: none; white-space: nowrap; }
 .cm-context-panel { width: min(100%, 38rem); margin-left: auto; padding: clamp(1.5rem, 4vw, 3rem); }
 .cm-context-panel__rail { display: flex; flex-wrap: wrap; align-items: center; gap: .65rem; margin-top: 2rem; color: rgb(var(--pc-text-rgb)); font: 600 .72rem/1.3 var(--font-mono), monospace; text-transform: uppercase; }
 .cm-context-panel__rail i { width: 1.5rem; height: 1px; background: rgb(var(--pc-blue-500-rgb)); }
diff --git a/src/components/cinematic/cinematic-media.css b/src/components/cinematic/cinematic-media.css
index bdbb0b1..6dc0df4 100644
--- a/src/components/cinematic/cinematic-media.css
+++ b/src/components/cinematic/cinematic-media.css
@@ -8,7 +8,12 @@
 .cm-video__canvas { opacity: 0; transition: opacity .5s ease; }
 .cm-video__canvas.is-visible { opacity: 1; }
 .cm-video__veil { background: linear-gradient(180deg, rgb(4 7 11 / .08), rgb(3 5 9 / .18) 58%, rgb(2 4 7 / .35)); }
-@media (prefers-reduced-motion: reduce) { .cm-video__canvas { display: none; } .cm-video__element { opacity: 1 !important; } }
+@media (prefers-reduced-motion: reduce) {
+  .cm-video__poster,
+  .cm-video__element { transition: none; }
+  .cm-video__element { display: block; }
+  .cm-video__canvas { display: none; }
+}
 
 .cm-experience.cm-marketing-experience .cm-video { background: rgb(181 200 210); }
 .cm-experience.cm-marketing-experience .cm-video__poster { background: radial-gradient(circle at 50% 42%, rgb(230 239 243) 0%, rgb(184 204 214) 48%, rgb(133 163 178) 100%); }
diff --git a/src/components/cinematic/cinematic-reference-final.css b/src/components/cinematic/cinematic-reference-final.css
index ff19047..dc4c2b9 100644
--- a/src/components/cinematic/cinematic-reference-final.css
+++ b/src/components/cinematic/cinematic-reference-final.css
@@ -183,7 +183,7 @@
 .cm-shell.cm-shell--core .cm-template-section-grid h2 {
   margin: 0;
   color: rgb(255 255 255);
-  font-family: var(--font-sans), Inter, sans-serif;
+  font-family: var(--font-brand), 'Golos Text', sans-serif;
   font-weight: 500;
   letter-spacing: -0.055em;
   line-height: 0.94;
@@ -276,8 +276,13 @@
   border-radius: 999px;
   font-size: 0.78rem;
   font-weight: 600;
+  white-space: nowrap;
+  flex: 0 0 auto;
 }
 
+.cm-shell.cm-shell--core .cm-nav__cta,
+.cm-shell.cm-shell--core .cm-template-context-card .cm-inline-link { white-space: nowrap; flex-shrink: 0; }
+
 .cm-shell.cm-shell--core .cm-actions__primary {
   border: 1px solid rgb(255 255 255 / 0.34);
   background: rgb(255 255 255 / 0.95);
diff --git a/src/components/funnels/ScrollVideoBackground.tsx b/src/components/funnels/ScrollVideoBackground.tsx
index 2815a70..a8ee63e 100644
--- a/src/components/funnels/ScrollVideoBackground.tsx
+++ b/src/components/funnels/ScrollVideoBackground.tsx
@@ -10,6 +10,8 @@ const BOOTSTRAP_FRAME_COUNT = 24
 const BOOTSTRAP_COLUMNS = 6
 const BOOTSTRAP_ROWS = 4
 const FIRST_FRAME_TIME_SECONDS = 0.04
+const MAX_CACHED_FRAMES = 72
+const MAX_CACHED_FRAME_WIDTH = 720
 
 type ScrollVideoBackgroundProps = {
   containerRef?: RefObject<HTMLElement | null>
@@ -45,15 +47,24 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
   const [bootstrapReady, setBootstrapReady] = useState(false)
   const [videoReady, setVideoReady] = useState(false)
   const [cacheReady, setCacheReady] = useState(false)
+  const [canvasPainted, setCanvasPainted] = useState(false)
+  const [prefersReducedMotion, setPrefersReducedMotion] = useState(false)
+
+  useEffect(() => {
+    const preference = window.matchMedia('(prefers-reduced-motion: reduce)')
+    const updatePreference = () => setPrefersReducedMotion(preference.matches)
+    updatePreference()
+    preference.addEventListener('change', updatePreference)
+    return () => preference.removeEventListener('change', updatePreference)
+  }, [])
 
   useEffect(() => {
     onProgressRef.current = onProgress
   }, [onProgress])
 
   useEffect(() => {
-    const reducedMotion = window.matchMedia('(prefers-reduced-motion: reduce)').matches
     const updateTarget = () => {
-      const result = reducedMotion
+      const result = prefersReducedMotion
         ? { progress: 0, isActive: true }
         : getProgress(containerRef?.current ?? null)
       targetRef.current = result.progress
@@ -61,7 +72,7 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
     }
 
     updateTarget()
-    if (reducedMotion) return
+    if (prefersReducedMotion) return
 
     window.addEventListener('scroll', updateTarget, { passive: true })
     window.addEventListener('resize', updateTarget)
@@ -69,7 +80,7 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
       window.removeEventListener('scroll', updateTarget)
       window.removeEventListener('resize', updateTarget)
     }
-  }, [containerRef])
+  }, [containerRef, prefersReducedMotion])
 
   useEffect(() => {
     const image = new Image()
@@ -162,8 +173,8 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
       })
       const duration = source.duration
       if (!Number.isFinite(duration) || duration <= 0) return
-      const count = Math.min(90, Math.max(24, Math.round(duration * 12)))
-      const width = Math.min(960, source.videoWidth || 960)
+      const count = Math.min(MAX_CACHED_FRAMES, Math.max(24, Math.round(duration * 10)))
+      const width = Math.min(MAX_CACHED_FRAME_WIDTH, source.videoWidth || MAX_CACHED_FRAME_WIDTH)
       const height = Math.max(1, Math.round(width * ((source.videoHeight || 540) / (source.videoWidth || 960))))
       const temp = document.createElement('canvas')
       temp.width = width
@@ -205,10 +216,11 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
   }, [videoReady])
 
   useEffect(() => {
-    const reducedMotion = window.matchMedia('(prefers-reduced-motion: reduce)').matches
-    if (reducedMotion) return
+    if (prefersReducedMotion) return
 
     let raf = 0
+    let previousTime = 0
+    let forceDraw = true
     const drawCover = (
       ctx: CanvasRenderingContext2D,
       source: CanvasImageSource,
@@ -218,11 +230,12 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
       height: number,
       sourceX = 0,
       sourceY = 0,
+      clear = true,
     ) => {
       const scale = Math.max(width / sourceWidth, height / sourceHeight)
       const drawWidth = sourceWidth * scale
       const drawHeight = sourceHeight * scale
-      ctx.clearRect(0, 0, width, height)
+      if (clear) ctx.clearRect(0, 0, width, height)
       ctx.drawImage(
         source,
         sourceX,
@@ -236,53 +249,110 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
       )
     }
 
-    const draw = () => {
-      smoothedRef.current += (targetRef.current - smoothedRef.current) * 0.12
+    const findFrame = (frames: Array<ImageBitmap | undefined>, start: number, direction: -1 | 1) => {
+      for (let index = start; index >= 0 && index < frames.length; index += direction) {
+        if (frames[index]) return { frame: frames[index]!, index }
+      }
+      return null
+    }
+
+    const draw = (time: number) => {
+      const delta = previousTime ? Math.min(64, time - previousTime) : 16.67
+      previousTime = time
+      const oldProgress = smoothedRef.current
+      smoothedRef.current += (targetRef.current - smoothedRef.current) * (1 - Math.exp(-delta / 88))
+      const moved = Math.abs(smoothedRef.current - oldProgress) > 0.00001
       const canvas = canvasRef.current
-      const video = videoRef.current
-      const canDrawCanvas = canvas && (cacheReady || bootstrapReady)
+      const canDrawCanvas = canvas && bootstrapReady
 
       if (canDrawCanvas) {
-        const dpr = Math.min(window.devicePixelRatio || 1, 2)
+        const dpr = Math.min(window.devicePixelRatio || 1, 1.5)
         const width = Math.round(window.innerWidth * dpr)
         const height = Math.round(window.innerHeight * dpr)
+        let resized = false
         if (canvas.width !== width || canvas.height !== height) {
           canvas.width = width
           canvas.height = height
+          resized = true
         }
         const ctx = canvas.getContext('2d')
 
-        if (ctx && cacheReady && frameCountRef.current > 0) {
-          const frames = framesRef.current
-          const targetIndex = Math.min(frameCountRef.current - 1, Math.round(smoothedRef.current * (frameCountRef.current - 1)))
-          let frame = frames[targetIndex]
-          if (!frame) {
-            for (let offset = 1; offset < frameCountRef.current; offset += 1) {
-              frame = frames[targetIndex - offset] ?? frames[targetIndex + offset]
-              if (frame) break
+        if (ctx && (forceDraw || resized || moved)) {
+          let drawn = false
+          try {
+            if (cacheReady && frameCountRef.current > 0) {
+              const frames = framesRef.current
+              const position = smoothedRef.current * (frameCountRef.current - 1)
+              const lower = findFrame(frames, Math.floor(position), -1)
+              const upper = findFrame(frames, Math.ceil(position), 1)
+
+              if (lower || upper) {
+                const first = lower ?? upper!
+                const second = upper ?? lower!
+                const blend = first.index === second.index
+                  ? 0
+                  : Math.min(1, Math.max(0, (position - first.index) / (second.index - first.index)))
+                drawCover(ctx, first.frame, first.frame.width, first.frame.height, width, height)
+                if (second.index !== first.index && blend > 0) {
+                  ctx.save()
+                  ctx.globalAlpha = blend
+                  drawCover(ctx, second.frame, second.frame.width, second.frame.height, width, height, 0, 0, false)
+                  ctx.restore()
+                }
+                drawn = true
+              }
             }
+
+            if (!drawn && bootstrapImageRef.current) {
+              const image = bootstrapImageRef.current
+              const framePosition = smoothedRef.current * (BOOTSTRAP_FRAME_COUNT - 1)
+              const lowerIndex = Math.floor(framePosition)
+              const upperIndex = Math.min(BOOTSTRAP_FRAME_COUNT - 1, lowerIndex + 1)
+              const blend = framePosition - lowerIndex
+              const sourceWidth = image.naturalWidth / BOOTSTRAP_COLUMNS
+              const sourceHeight = image.naturalHeight / BOOTSTRAP_ROWS
+              const drawBootstrap = (frameIndex: number, clear: boolean, alpha = 1) => {
+                ctx.save()
+                ctx.globalAlpha = alpha
+                drawCover(
+                  ctx,
+                  image,
+                  sourceWidth,
+                  sourceHeight,
+                  width,
+                  height,
+                  (frameIndex % BOOTSTRAP_COLUMNS) * sourceWidth,
+                  Math.floor(frameIndex / BOOTSTRAP_COLUMNS) * sourceHeight,
+                  clear,
+                )
+                ctx.restore()
+              }
+              drawBootstrap(lowerIndex, true)
+              if (upperIndex !== lowerIndex && blend > 0) drawBootstrap(upperIndex, false, blend)
+              drawn = true
+            }
+          } catch {
+            // Keep the last successfully painted canvas frame if a source is transiently unavailable.
+          }
+
+          if (drawn) {
+            forceDraw = false
+            if (!canvasPainted) setCanvasPainted(true)
           }
-          if (frame) drawCover(ctx, frame, frame.width, frame.height, width, height)
-        } else if (ctx && bootstrapReady && bootstrapImageRef.current) {
-          const image = bootstrapImageRef.current
-          const sourceWidth = image.naturalWidth / BOOTSTRAP_COLUMNS
-          const sourceHeight = image.naturalHeight / BOOTSTRAP_ROWS
-          const frameIndex = Math.min(BOOTSTRAP_FRAME_COUNT - 1, Math.round(smoothedRef.current * (BOOTSTRAP_FRAME_COUNT - 1)))
-          const sourceX = (frameIndex % BOOTSTRAP_COLUMNS) * sourceWidth
-          const sourceY = Math.floor(frameIndex / BOOTSTRAP_COLUMNS) * sourceHeight
-          drawCover(ctx, image, sourceWidth, sourceHeight, width, height, sourceX, sourceY)
         }
-      } else if (videoReady && video && Number.isFinite(video.duration)) {
-        const wanted = smoothedRef.current * Math.max(0, video.duration - 0.05)
-        if (!video.seeking && Math.abs(video.currentTime - wanted) > 0.04) video.currentTime = wanted
       }
-      raf = requestAnimationFrame(draw)
+
+      if (!prefersReducedMotion || !canvasPainted && bootstrapReady) {
+        raf = requestAnimationFrame(draw)
+      }
     }
 
     raf = requestAnimationFrame(draw)
     return () => cancelAnimationFrame(raf)
-  }, [bootstrapReady, cacheReady, videoReady])
+  }, [bootstrapReady, cacheReady, canvasPainted, prefersReducedMotion])
 
-  const canvasVisible = bootstrapReady || cacheReady
-  return <><link rel="preload" as="image" href={BOOTSTRAP_SPRITE_URL} /><div className={`cm-video ${className}`.trim()} aria-hidden="true"><div className={`cm-video__poster ${bootstrapReady || videoReady || cacheReady ? 'is-hidden' : ''}`} /><video ref={videoRef} className={`cm-video__element ${canvasVisible || cacheReady ? 'is-hidden' : videoReady ? 'is-visible' : ''}`} src={VIDEO_URL} muted playsInline preload="metadata" /><canvas ref={canvasRef} className={`cm-video__canvas ${canvasVisible ? 'is-visible' : ''}`} data-frame-cache-ready={cacheReady ? 'true' : 'false'} /><div className="cm-video__veil" /></div></>
+  const canvasVisible = canvasPainted && !prefersReducedMotion
+  const videoHidden = canvasPainted || prefersReducedMotion
+  const renderState = cacheReady ? 'cache' : canvasPainted ? 'bootstrap' : 'poster'
+  return <><link rel="preload" as="image" href={BOOTSTRAP_SPRITE_URL} /><div className={`cm-video ${className}`.trim()} aria-hidden="true" data-render-state={renderState}><div className={`cm-video__poster ${canvasVisible ? 'is-hidden' : ''}`} /><video ref={videoRef} className={`cm-video__element ${videoHidden ? 'is-hidden' : videoReady ? 'is-visible' : ''}`} src={VIDEO_URL} muted playsInline preload="metadata" /><canvas ref={canvasRef} className={`cm-video__canvas ${canvasVisible ? 'is-visible' : ''}`} data-frame-cache-ready={cacheReady ? 'true' : 'false'} /><div className="cm-video__veil" /></div></>
 }
diff --git a/tests/evidence/cinematic-final-quality.spec.ts b/tests/evidence/cinematic-final-quality.spec.ts
new file mode 100644
index 0000000..c8d771e
--- /dev/null
+++ b/tests/evidence/cinematic-final-quality.spec.ts
@@ -0,0 +1,211 @@
+import { expect, test, type Page } from '@playwright/test'
+
+const baseUrl = process.env.WAVE1_BASE_URL
+if (!baseUrl) throw new Error('WAVE1_BASE_URL is required')
+
+const CORE_ROUTES = ['/', '/evermind', '/nevermind', '/mastermind'] as const
+const VIEWPORTS = [1440, 1280, 1024, 768, 430, 390, 360] as const
+
+function routeUrl(route: string) {
+  return new URL(route, baseUrl).toString()
+}
+
+async function sampleRenderedCanvas(page: Page) {
+  return page.evaluate(async () => {
+    const canvas = document.querySelector<HTMLCanvasElement>('.cm-video__canvas')
+    if (!canvas) throw new Error('Cinematic canvas is missing')
+
+    const scratch = document.createElement('canvas')
+    scratch.width = 48
+    scratch.height = 27
+    const context = scratch.getContext('2d')
+    if (!context) throw new Error('Canvas 2D context is unavailable')
+
+    const result = { samples: 0, uncoveredSamples: 0, visibleCanvasSamples: 0, states: new Set<string>() }
+    const read = () => {
+      const root = canvas.closest<HTMLElement>('.cm-video')
+      const poster = root?.querySelector<HTMLElement>('.cm-video__poster')
+      result.states.add(root?.dataset.renderState ?? 'missing')
+      const canvasOpacity = Number.parseFloat(getComputedStyle(canvas).opacity)
+      const posterOpacity = Number.parseFloat(poster ? getComputedStyle(poster).opacity : '0')
+      if (canvasOpacity >= 0.99) result.visibleCanvasSamples += 1
+      context.clearRect(0, 0, scratch.width, scratch.height)
+      context.drawImage(
+        canvas,
+        Math.max(0, (canvas.width - scratch.width) / 2),
+        Math.max(0, (canvas.height - scratch.height) / 2),
+        Math.min(canvas.width, scratch.width),
+        Math.min(canvas.height, scratch.height),
+        0,
+        0,
+        scratch.width,
+        scratch.height,
+      )
+      const pixels = context.getImageData(0, 0, scratch.width, scratch.height).data
+      let alpha = 0
+      for (let index = 3; index < pixels.length; index += 4) alpha += pixels[index]
+      result.samples += 1
+      // The canvas is intentionally transparent while the poster is the visible
+      // fallback. Count a gap only when neither layer covers the transition.
+      if (canvasOpacity + posterOpacity < 0.98 || (canvasOpacity >= 0.99 && alpha === 0)) {
+        result.uncoveredSamples += 1
+      }
+    }
+
+    const step = async (direction: -1 | 1, count: number, distance: number) => {
+      for (let index = 0; index < count; index += 1) {
+        await new Promise<void>(resolve => requestAnimationFrame(() => resolve()))
+        window.scrollBy(0, direction * distance)
+        read()
+      }
+    }
+
+    for (let cycle = 0; cycle < 3; cycle += 1) {
+      await step(1, 42, 5)
+      await step(-1, 42, 5)
+      await step(1, 28, 26)
+      await step(-1, 28, 26)
+    }
+
+    return { ...result, states: [...result.states] }
+  })
+}
+
+test.describe('ProChat final public-site quality', () => {
+  test('core routes keep every cinematic background painted through repeated scroll reversals', async ({ page }) => {
+    test.setTimeout(120_000)
+    await page.setViewportSize({ width: 1440, height: 900 })
+    await page.emulateMedia({ reducedMotion: 'no-preference' })
+
+    for (const route of CORE_ROUTES) {
+      const response = await page.goto(routeUrl(route), { waitUntil: 'domcontentloaded' })
+      expect(response?.status(), `${route} response`).toBe(200)
+      await expect(page.locator('.cm-video__canvas.is-visible')).toBeVisible()
+      await expect.poll(() => page.locator('.cm-video__canvas').evaluate(canvas => getComputedStyle(canvas).opacity)).toBe('1')
+
+      const sampled = await sampleRenderedCanvas(page)
+      expect(sampled.samples, `${route} sampled frame count`).toBeGreaterThan(200)
+      expect(sampled.visibleCanvasSamples, `${route} visible canvas samples`).toBeGreaterThan(200)
+      expect(sampled.uncoveredSamples, `${route} uncovered background samples`).toBe(0)
+      expect(sampled.states.every(state => state === 'bootstrap' || state === 'cache'), `${route} media state stays painted`).toBe(true)
+    }
+  })
+
+  test('four cinematic pages share one clean header and CTA labels stay on a single line at all target widths', async ({ page }) => {
+    test.setTimeout(180_000)
+    await page.emulateMedia({ reducedMotion: 'no-preference' })
+
+    for (const width of VIEWPORTS) {
+      await page.setViewportSize({ width, height: width < 600 ? 844 : 900 })
+      for (const route of CORE_ROUTES) {
+        const response = await page.goto(routeUrl(route), { waitUntil: 'domcontentloaded' })
+        expect(response?.status(), `${route} at ${width}px`).toBe(200)
+
+        const evidence = await page.evaluate(() => {
+          const heading = document.querySelector<HTMLElement>('main h1')
+          const ctas = Array.from(document.querySelectorAll<HTMLElement>(
+            '.cm-actions a, .cm-nav__cta',
+          )).filter(element => element.getBoundingClientRect().width > 0).map(element => {
+            const textNode = Array.from(element.childNodes).find(node => node.nodeType === Node.TEXT_NODE && node.textContent?.trim())
+            let lineCount = 0
+            if (textNode) {
+              const range = document.createRange()
+              range.selectNode(textNode)
+              lineCount = new Set(Array.from(range.getClientRects()).map(rect => Math.round(rect.top))).size
+            }
+            return {
+              label: element.textContent?.trim() ?? '',
+              whiteSpace: getComputedStyle(element).whiteSpace,
+              lines: lineCount,
+              clipped: element.scrollWidth > element.clientWidth + 1,
+            }
+          })
+
+          return {
+            headers: document.querySelectorAll('header').length,
+            navs: document.querySelectorAll('nav[aria-label="Primary navigation"]').length,
+            mains: document.querySelectorAll('main').length,
+            footers: document.querySelectorAll('footer').length,
+            overflow: document.documentElement.scrollWidth > window.innerWidth,
+            headingFont: heading ? getComputedStyle(heading).fontFamily : '',
+            brandFont: heading ? getComputedStyle(heading).fontFamily.includes('Golos_Text') || getComputedStyle(heading).fontFamily.includes('Golos Text') : false,
+            ctas,
+          }
+        })
+
+        expect(evidence.headers, `${route} header count`).toBe(1)
+        expect(evidence.navs, `${route} primary nav count`).toBe(1)
+        expect(evidence.mains, `${route} main count`).toBe(1)
+        expect(evidence.footers, `${route} footer count`).toBe(0)
+        expect(evidence.overflow, `${route} overflow at ${width}px`).toBe(false)
+        expect(evidence.brandFont, `${route} H1 uses Golos Text (${evidence.headingFont})`).toBe(true)
+        expect(evidence.ctas.every(cta => cta.whiteSpace === 'nowrap' && cta.lines <= 1 && !cta.clipped), `${route} CTA labels at ${width}px: ${JSON.stringify(evidence.ctas)}`).toBe(true)
+      }
+    }
+  })
+
+  test('reduced motion leaves core stages readable without video playback or canvas painting', async ({ page }) => {
+    await page.setViewportSize({ width: 1440, height: 900 })
+    await page.emulateMedia({ reducedMotion: 'reduce' })
+
+    for (const route of CORE_ROUTES) {
+      await page.goto(routeUrl(route), { waitUntil: 'domcontentloaded' })
+      await expect(page.locator('main h1')).toBeVisible()
+      const evidence = await page.evaluate(() => ({
+        reduced: matchMedia('(prefers-reduced-motion: reduce)').matches,
+        chaptersVisible: Array.from(document.querySelectorAll<HTMLElement>('[data-cinematic-chapter]')).every(element => {
+          const style = getComputedStyle(element)
+          return style.opacity === '1' && style.visibility !== 'hidden' && element.getBoundingClientRect().height > 0
+        }),
+        videoDisplay: document.querySelector('.cm-video__element') ? getComputedStyle(document.querySelector('.cm-video__element') as HTMLElement).display : 'not-present',
+        canvasDisplay: document.querySelector('.cm-video__canvas') ? getComputedStyle(document.querySelector('.cm-video__canvas') as HTMLElement).display : 'not-present',
+        videoPaused: document.querySelector('.cm-video__element') ? (document.querySelector('.cm-video__element') as HTMLVideoElement).paused : true,
+        navLinkCount: Array.from(document.querySelectorAll<HTMLAnchorElement>('nav[aria-label="Primary navigation"] a')).filter(link => link.getBoundingClientRect().width > 0).length,
+      }))
+      expect(evidence.reduced).toBe(true)
+      expect(evidence.chaptersVisible, `${route} readable content`).toBe(true)
+      expect(evidence.canvasDisplay, `${route} reduced-motion canvas layer`).toBe('none')
+      expect(evidence.videoPaused, `${route} reduced-motion video must remain static`).toBe(true)
+      expect(evidence.navLinkCount, `${route} usable navigation`).toBeGreaterThan(0)
+    }
+  })
+
+  test('switching to reduced motion after load restores one stable poster layer', async ({ page }) => {
+    await page.setViewportSize({ width: 1440, height: 900 })
+    await page.emulateMedia({ reducedMotion: 'no-preference' })
+    await page.goto(routeUrl('/'), { waitUntil: 'domcontentloaded' })
+    await expect(page.locator('.cm-video__canvas.is-visible')).toBeVisible()
+
+    await page.emulateMedia({ reducedMotion: 'reduce' })
+    await expect.poll(() => page.locator('.cm-video__poster').evaluate(element => getComputedStyle(element).opacity)).toBe('1')
+
+    const evidence = await page.evaluate(() => ({
+      canvasDisplay: getComputedStyle(document.querySelector('.cm-video__canvas') as HTMLElement).display,
+      videoOpacity: getComputedStyle(document.querySelector('.cm-video__element') as HTMLElement).opacity,
+      posterOpacity: getComputedStyle(document.querySelector('.cm-video__poster') as HTMLElement).opacity,
+    }))
+    expect(evidence).toEqual({ canvasDisplay: 'none', videoOpacity: '0', posterOpacity: '1' })
+
+    await page.emulateMedia({ reducedMotion: 'no-preference' })
+    await expect(page.locator('.cm-video__canvas.is-visible')).toBeVisible()
+    const firstFrame = await page.locator('.cm-video__canvas').evaluate(element => {
+      const canvas = element as HTMLCanvasElement
+      const sample = document.createElement('canvas')
+      sample.width = 16
+      sample.height = 9
+      const context = sample.getContext('2d')!
+      context.drawImage(canvas, 0, 0, sample.width, sample.height)
+      return [...context.getImageData(0, 0, sample.width, sample.height).data].reduce((sum, value, index) => sum + value * (index + 1), 0)
+    })
+    await page.evaluate(() => window.scrollTo({ top: document.documentElement.scrollHeight, behavior: 'instant' }))
+    await expect.poll(() => page.locator('.cm-video__canvas').evaluate(element => {
+      const canvas = element as HTMLCanvasElement
+      const sample = document.createElement('canvas')
+      sample.width = 16
+      sample.height = 9
+      const context = sample.getContext('2d')!
+      context.drawImage(canvas, 0, 0, sample.width, sample.height)
+      return [...context.getImageData(0, 0, sample.width, sample.height).data].reduce((sum, value, index) => sum + value * (index + 1), 0)
+    }), { timeout: 10_000 }).not.toBe(firstFrame)
+  })
+})
diff --git a/tests/evidence/cinematic-marketing-experience.spec.ts b/tests/evidence/cinematic-marketing-experience.spec.ts
index 68741b3..fea5f1f 100644
--- a/tests/evidence/cinematic-marketing-experience.spec.ts
+++ b/tests/evidence/cinematic-marketing-experience.spec.ts
@@ -323,13 +323,14 @@ test.describe('cinematic marketing experience', () => {
         })),
         canvasDisplay: getComputedStyle(document.querySelector('.cm-video__canvas') as HTMLElement).display,
         videoOpacity: getComputedStyle(document.querySelector('.cm-video__element') as HTMLElement).opacity,
+        posterOpacity: getComputedStyle(document.querySelector('.cm-video__poster') as HTMLElement).opacity,
         videoCurrentTime: (document.querySelector('.cm-video__element') as HTMLVideoElement).currentTime,
         links: Array.from(document.querySelectorAll<HTMLAnchorElement>('main a')).filter(link => link.getBoundingClientRect().width > 0).length,
       }))
       expect(evidence.reduced).toBe(true)
       expect(evidence.chapters.every(chapter => chapter.visible && chapter.rect > 0), `${route} readable chapters`).toBe(true)
       expect(evidence.canvasDisplay, `${route} canvas is not required under reduced motion`).toBe('none')
-      expect(evidence.videoOpacity, `${route} keeps one stable media layer under reduced motion`).toBe('1')
+      expect(evidence.videoOpacity === '1' || evidence.posterOpacity === '1', `${route} keeps a stable media frame or poster under reduced motion`).toBe(true)
       expect(evidence.links, `${route} links remain usable under reduced motion`).toBeGreaterThan(0)
       await page.evaluate(() => window.scrollTo({ top: document.documentElement.scrollHeight, behavior: 'instant' }))
       await page.waitForTimeout(250)

From b6a91195ad3bdbd6a92b04cf2b65221605550395 Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Tue, 29 Sep 2026 11:27:33 +0100
Subject: [PATCH 3/9] fix: finalize cinematic ProChat public site

(cherry picked from commit e6c9e17c04d8b87fdd3820f4612ee3c70395e21c)
---
 README.md                                     |  11 +-
 docs/ai-guidelines.md                         |   4 +-
 docs/content-platform.md                      |  12 +-
 docs/design-system.md                         |   4 +-
 docs/design/MOTION_STORYBOARD.md              |   4 +-
 docs/development.md                           |   8 +-
 docs/docs-automation.md                       |   2 +-
 docs/getting-started.md                       |   6 +-
 docs/homepage-design-orchestration.md         |   4 +-
 docs/homepage-design-spec.md                  |   6 +-
 docs/homepage-technical-design.md             |   4 +-
 docs/homepage-validation-plan.md              |  10 +-
 docs/implementation-plan.md                   |  17 +-
 docs/overview.md                              |   8 +-
 docs/production-lifecycle.md                  |   8 +-
 docs/repo-status.md                           |   8 +-
 docs/roadmap.md                               |  19 +-
 docs/token-architecture.md                    |   4 +-
 docs/website-copy-blueprint.md                |   4 +-
 .../components/layout/PublicLegalPage.tsx     |  11 +-
 .../(marketing)/contact/ContactPageClient.tsx |  13 +-
 .../(marketing)/contact/ContactPageMarkup.tsx |  76 ++--
 src/app/docs/layout.tsx                       |  11 +-
 src/app/docs/page.tsx                         |  27 +-
 src/components/CinematicMarketingShell.tsx    |   7 +-
 .../cinematic/cinematic-utility.css           | 334 ++++++++++++++++++
 .../funnels/ScrollVideoBackground.tsx         |  64 ++--
 src/helpers/public-route-design.ts            |   8 +-
 tests/evidence/canonical-chrome-proof.spec.ts |  36 +-
 tests/evidence/canonical-route-smoke.spec.ts  |  29 +-
 .../evidence/cinematic-final-quality.spec.ts  | 227 +++++++++++-
 tests/evidence/public-v4-route-audit.spec.ts  |  39 +-
 32 files changed, 798 insertions(+), 227 deletions(-)
 create mode 100644 src/components/cinematic/cinematic-utility.css

diff --git a/README.md b/README.md
index 99cd653..95bfb4f 100644
--- a/README.md
+++ b/README.md
@@ -67,13 +67,10 @@ Use these docs for website implementation only:
 5. `docs/design/` — design principles, visual language, product visual library, copy-to-visual mapping, motion storyboards, design-lab rules, and component contracts.
 6. `docs/platform/` — page architecture, responsive strategy, accessibility strategy, and performance strategy.
 7. `docs/migration/` — foundational legacy sweep, migration matrix, and content, route, component, style, motion, asset, and dependency audits.
-8. `docs/homepage-design-spec.md` — central homepage design brief and closed decisions.
-9. `docs/homepage-visual-storyboard.md` and `docs/homepage-example-data.md` — visual states and sanitized prototype content.
-10. `docs/homepage-technical-design.md` and `docs/homepage-design-orchestration.md` — historical build-planning evidence; check current source before relying on details.
-11. `docs/homepage-validation-plan.md` — visual, motion, accessibility, performance, and release gates.
-12. `docs/roadmap.md` — canonical 15-phase public-platform roadmap and continuous-governance status.
-13. `docs/implementation-plan.md` — dependency-gated implementation tasks with validation, rollback, and commit boundaries.
-14. `docs/auth-status.md`, `docs/deployment.md`, `docs/content-platform.md`, and related implementation docs — runtime and deployment facts.
+8. `docs/homepage-validation-plan.md` — active visual, motion, accessibility, performance, and release checks.
+9. `docs/implementation-plan.md` and `docs/roadmap.md` — current website implementation status and active follow-up.
+10. `docs/homepage-design-spec.md`, `docs/homepage-visual-storyboard.md`, `docs/homepage-example-data.md`, `docs/homepage-technical-design.md`, and `docs/homepage-design-orchestration.md` — historical/pre-build evidence; check current source before relying on details.
+11. `docs/auth-status.md`, `docs/deployment.md`, `docs/content-platform.md`, and related implementation docs — runtime and deployment facts.
 
 ## Architecture overview
 
diff --git a/docs/ai-guidelines.md b/docs/ai-guidelines.md
index f08b4b7..0c7ed58 100644
--- a/docs/ai-guidelines.md
+++ b/docs/ai-guidelines.md
@@ -4,11 +4,11 @@ These rules describe how AI-assisted development and internal automation may int
 
 ## Current product boundary
 
-The canonical public products are Memory, Memory for QA, and Workbench. The public site also exposes Docs, Contact, Privacy, and Terms.
+Evermind and Nevermind are the connected Infinite Brain product system; Mastermind is a separate ProChat product. ProChat Memory for QA is paused/historical. The public site also exposes Docs, Contact, Privacy, and Terms.
 
 Retired Kits, ProChat OS, AI Workflows, Studio, Proof, Prompts, generated Docs, checkout/licensing, and MailerLite systems are historical only. Do not restore them because older docs or Git history mention them.
 
-BuildFlow may appear only where a technical/internal compatibility identifier is required by Workbench implementation contracts.
+BuildFlow may appear only where a technical/internal compatibility identifier is required by Mastermind implementation contracts.
 
 ## Repository safety
 
diff --git a/docs/content-platform.md b/docs/content-platform.md
index e6ef24e..a21c4ff 100644
--- a/docs/content-platform.md
+++ b/docs/content-platform.md
@@ -7,9 +7,9 @@ This document describes the current root-domain public content and SEO contract.
 The active canonical public surface is intentionally small:
 
 - `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
+- `/evermind`
+- `/nevermind`
+- `/mastermind`
 - `/docs`
 - `/contact`
 - `/privacy`
@@ -19,7 +19,7 @@ The root sitemap emits only those eight routes.
 
 ## Documentation
 
-The active `/docs` route is a repository-oriented hub for Memory for QA and Workbench. It links readers to repository documentation, issues, and beta/contact paths.
+The active `/docs` route is a reading-friendly hub for Evermind, Nevermind, and Mastermind. It links readers to product routes and contact; canonical product documentation remains with the relevant repositories.
 
 The former generated public documentation system is retired. In particular:
 
@@ -45,11 +45,11 @@ Current SEO behavior is deliberately bounded:
 - compatibility aliases redirect into canonical public routes rather than publishing duplicate bodies
 - retired product/content routes are not emitted by the sitemap
 
-Structured data and page metadata should describe only active Memory, Memory for QA, Workbench, Docs, Contact, and legal surfaces.
+Structured data and page metadata should describe only active Evermind, Nevermind, Mastermind, Docs, Contact, and legal surfaces.
 
 ## Compatibility boundaries
 
-Some historical URLs remain only as redirect compatibility, including BuildFlow / ProChat OS aliases to Workbench, Learn aliases to Docs, legal aliases to the current Privacy/Terms pages, and waitlist aliases into the Memory for QA beta contact flow.
+Some historical URLs remain only as redirect compatibility, including BuildFlow / ProChat OS aliases to Mastermind, Learn aliases to Docs, legal aliases to the current Privacy/Terms pages, and retired waitlist aliases into Contact.
 
 Compatibility routes must not restore retired product bodies or duplicate canonical content.
 
diff --git a/docs/design-system.md b/docs/design-system.md
index cd5d36d..91df2dd 100644
--- a/docs/design-system.md
+++ b/docs/design-system.md
@@ -1,11 +1,13 @@
 # ProChat Design-System Implementation Bridge
 
-**Status:** repository-local implementation guidance  
+**Status:** historical design-system migration guidance
 **Canonical design truth:** `DESIGN.md` and `brand-spec.md`  
 **Canonical strategy and company design:** Mind
 
 This file explains how the approved ProChat design system should be introduced into the existing website repository.
 
+> This migration plan predates the current cinematic and utility route systems. Its planned token files, GSAP direction, and migration sequence are not current implementation facts. For the live route contract and stylesheet ownership, use `docs/overview.md`, `docs/implementation-plan.md`, and `docs/homepage-validation-plan.md`.
+
 Existing tokens, themes, gradients, fonts, hero components, and page-specific styles remain implementation inventory. They are not canonical design authority and must not be extended by default.
 
 ## Required reading order
diff --git a/docs/design/MOTION_STORYBOARD.md b/docs/design/MOTION_STORYBOARD.md
index 50ca1de..900545f 100644
--- a/docs/design/MOTION_STORYBOARD.md
+++ b/docs/design/MOTION_STORYBOARD.md
@@ -1,9 +1,9 @@
 # ProChat Motion Storyboard
 
-**Status:** canonical motion planning document  
+**Status:** historical pre-build motion planning document
 **Authority:** `docs/homepage-visual-storyboard.md`, `DESIGN.md`
 
-> Historical pre-cinematic storyboard. Its Memory/QA-specific sequences are not the current public route specification. Current routes `/`, `/evermind`, `/nevermind`, and `/mastermind` share two full-viewport sections, an intentional 80vh empty spacer, and one scroll-driven cinematic media layer. Other public routes are outside this visual redesign. See `docs/homepage-validation-plan.md` and `docs/implementation-plan.md` for active behavior and evidence.
+> Historical pre-build storyboard. Its Memory/QA-specific sequences are not the current public route specification. Current routes `/`, `/evermind`, `/nevermind`, and `/mastermind` share three ordered product stages, one scroll-driven cinematic media layer, and shared geometry. `/docs`, `/contact`, `/privacy`, and `/terms` use the static utility shell; legal body copy remains unchanged. See `docs/homepage-validation-plan.md` and `docs/implementation-plan.md` for active behavior and evidence.
 
 ## Purpose
 
diff --git a/docs/development.md b/docs/development.md
index b433fe9..03b6f35 100644
--- a/docs/development.md
+++ b/docs/development.md
@@ -27,9 +27,9 @@ npm run build
 The active public website is exactly:
 
 - `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
+- `/evermind`
+- `/nevermind`
+- `/mastermind`
 - `/docs`
 - `/contact`
 - `/privacy`
@@ -39,7 +39,7 @@ Compatibility aliases may redirect into those routes, but new public work should
 
 ## Documentation workflow
 
-Internal repository documentation lives under `docs/` and `docs-public/`. The active public `/docs` page is a lean repository hub for Memory for QA and Workbench.
+Internal repository documentation lives under `docs/` and `docs-public/`. The active public `/docs` page is a reading-friendly hub for Evermind, Nevermind, and Mastermind, presented in the shared static utility shell. Contact and legal routes use that shell too; legal body text remains unchanged.
 
 The former Nextra/generated `src/content/docs` + `scripts/docs` pipeline is retired.
 
diff --git a/docs/docs-automation.md b/docs/docs-automation.md
index a7c1fe9..3898506 100644
--- a/docs/docs-automation.md
+++ b/docs/docs-automation.md
@@ -15,7 +15,7 @@ The cleanup removed the generated documentation system, including:
 
 ## Current documentation model
 
-The active public documentation surface is the lean `/docs` repository hub for Memory for QA and Workbench. Canonical product documentation should live with the relevant repositories and be linked from that hub.
+The active public documentation surface is a reading-friendly `/docs` hub for Evermind, Nevermind, and Mastermind. Canonical product documentation should live with the relevant repositories and be linked from that hub.
 
 Internal repository documentation remains under `docs/` and `docs-public/` as appropriate.
 
diff --git a/docs/getting-started.md b/docs/getting-started.md
index c269518..9aeb80f 100644
--- a/docs/getting-started.md
+++ b/docs/getting-started.md
@@ -29,9 +29,9 @@ The default local port is `3056` unless overridden by `PORT`.
 Canonical public routes:
 
 - `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
+- `/evermind`
+- `/nevermind`
+- `/mastermind`
 - `/docs`
 - `/contact`
 - `/privacy`
diff --git a/docs/homepage-design-orchestration.md b/docs/homepage-design-orchestration.md
index 5641af3..e3632ea 100644
--- a/docs/homepage-design-orchestration.md
+++ b/docs/homepage-design-orchestration.md
@@ -1,9 +1,11 @@
 # ProChat Homepage Design Orchestration
 
-**Status:** approved pre-build workflow  
+**Status:** historical pre-build workflow
 **Source skills:** Brain `ai/skills`  
 **Scope:** design research, specification, prototyping, implementation, review, and polish
 
+> This workflow is retained as historical planning evidence, not as current release authority. Follow the active source and validation references in `docs/overview.md`, `docs/implementation-plan.md`, and `docs/homepage-validation-plan.md`.
+
 This document defines how GPT-5.6 Sol uses the Brain design skills across research, specification, prototyping, implementation, review, and polish. It prevents duplicated design authority, conflicting methods, and unstructured execution.
 
 ## Operating principle
diff --git a/docs/homepage-design-spec.md b/docs/homepage-design-spec.md
index 1030be7..d5559ea 100644
--- a/docs/homepage-design-spec.md
+++ b/docs/homepage-design-spec.md
@@ -1,13 +1,15 @@
 # ProChat Homepage Design Specification
 
-**Status:** approved pre-build design brief  
+**Status:** superseded pre-build design brief; retained as historical evidence
 **Canonical copy:** Mind `canonical-homepage-copy.md`  
 **Design system:** `DESIGN.md` and `brand-spec.md`  
 **Visual specification:** `docs/homepage-visual-storyboard.md`  
 **Technical architecture:** `docs/homepage-technical-design.md`  
 **Skill workflow:** `docs/homepage-design-orchestration.md`
 
-This is the central repository-local brief for the ProChat homepage. It explains what must be designed, why it matters, and which decisions are already closed.
+> This specification describes the retired Memory-first/Memory-for-QA/Workbench homepage direction. It is not current product, copy, route, or acceptance authority. Current direction is defined by Mind and the shipped implementation; use `docs/homepage-validation-plan.md`, `docs/implementation-plan.md`, and `docs/roadmap.md` for current website state.
+
+This document preserves the original pre-build design rationale and closed decisions for historical reference only.
 
 ## Objective
 
diff --git a/docs/homepage-technical-design.md b/docs/homepage-technical-design.md
index 1d04e58..11d1a76 100644
--- a/docs/homepage-technical-design.md
+++ b/docs/homepage-technical-design.md
@@ -1,9 +1,11 @@
 # ProChat Homepage Technical Design
 
-**Status:** approved pre-build architecture  
+**Status:** superseded pre-build architecture; retained as historical evidence
 **Scope:** homepage design system, prototypes, implementation, motion, validation, and production hardening  
 **Does not authorize:** production homepage implementation before prototype and plan-review gates pass
 
+> This was a pre-build plan and does not describe the current renderer or authorize further implementation. Current source and acceptance authority are `docs/implementation-plan.md`, `docs/homepage-validation-plan.md`, and `docs/roadmap.md`.
+
 ## Current repository stack
 
 Verified current stack:
diff --git a/docs/homepage-validation-plan.md b/docs/homepage-validation-plan.md
index 7e61425..0e7b3ee 100644
--- a/docs/homepage-validation-plan.md
+++ b/docs/homepage-validation-plan.md
@@ -1,7 +1,7 @@
 # ProChat public-site validation plan
 
 **Status:** current validation contract for the cinematic public website
-**Applies visually to:** `/`, `/evermind`, `/nevermind`, and `/mastermind`. Other canonical routes are regression-only.
+**Applies visually to:** `/`, `/evermind`, `/nevermind`, `/mastermind`, `/contact`, and `/docs`. `/privacy` and `/terms` share the static utility shell; their legal body meaning remains outside visual copy changes.
 
 Product truth comes from `mind/organizations/prochat/brand/`; implementation state comes from current source and `docs/implementation-plan.md`. Older homepage/page-architecture plans are historical unless explicitly reconciled.
 
@@ -10,7 +10,7 @@ Product truth comes from `mind/organizations/prochat/brand/`; implementation sta
 - Evermind is free, open-source, human-owned memory; Nevermind is its paid capability/convenience layer; Mastermind is a separate free planning/orchestration product.
 - ProChat Memory for QA is paused/historical, not a current public product.
 - Home communicates the product family accurately; each product route explains its own role and CTA destination.
-- Product identities on the four cinematic routes match Mind; do not change Contact/Docs copy or navigation in this release.
+- Product identities on all active routes match Mind. Contact and Docs use Evermind/Nevermind/Mastermind terminology; legal shell changes must not rewrite legal body copy.
 - Claims about privacy, security, integrations, pricing, availability, and outcomes match Mind and verified implementation. Privacy/Terms meaning must not be changed by a shell migration.
 
 ## Shared cinematic-page checks
@@ -23,7 +23,7 @@ For `/`, `/evermind`, `/nevermind`, and `/mastermind` verify:
 - shared route data changes copy without changing geometry or visual system;
 - capability display is one translucent glass panel, with background media visibly present through it;
 - media visibly scrubs with scroll, forward and reverse; adjacent frames blend and renderer handoffs stay continuously painted;
-- two stages activate in reading order, without text overlap or jarring chapter jumps;
+- the Evermind / Memory, Nevermind / Context, and Mastermind / Execution stages activate in reading order, without text overlap or jarring chapter jumps;
 - every CTA is usable and remains on one line without clipping.
 
 ## Responsive and motion matrix
@@ -32,6 +32,10 @@ Check widths `1440`, `1280`, `1024`, `768`, `430`, `390`, and `360` pixels. Revi
 
 Under `prefers-reduced-motion: reduce`, verify all stages and links remain available, no scroll scrubbing occurs, media freezes to one stable poster/frame, video and canvas are not visibly duplicated, and layout geometry remains understandable. Include a reduced-motion desktop screenshot.
 
+## Static utility shell
+
+For `/contact`, `/docs`, `/privacy`, and `/terms`, verify one transparent full-width navigation, no legacy footer or dark capsule, no video/canvas animation, Golos Text display headings, no horizontal overflow, and readable glass surfaces. Verify Contact submission and mobile navigation; verify Docs content remains readable at desktop, laptop, tablet, and mobile widths. Privacy and Terms body copy must remain semantically unchanged.
+
 ## Accessibility, performance, and runtime
 
 - Keyboard, focus-visible, accessible names/landmarks, form labels/errors, and target sizes pass.
diff --git a/docs/implementation-plan.md b/docs/implementation-plan.md
index 8025ae0..f0b1053 100644
--- a/docs/implementation-plan.md
+++ b/docs/implementation-plan.md
@@ -1,6 +1,6 @@
 # ProChat website implementation plan
 
-Status: cinematic/site quality implementation and release validation in progress.
+Status: cinematic renderer, utility shell, and local validation complete; commit/push and production release remain pending.
 
 Last reviewed: 2026-09-29.
 
@@ -8,32 +8,33 @@ Last reviewed: 2026-09-29.
 
 Canonical public routes: `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, `/contact`, `/privacy`, and `/terms`. Product truth is governed by Mind at `mind/organizations/prochat/brand/`; the current identities are Evermind, Nevermind, and Mastermind. ProChat Memory for QA is paused/historical.
 
-The first four routes use a shared cinematic implementation: `HomeV2` or `CinematicProductFunnel` supplies page data to `CinematicMarketingPage`; the shared page uses `CinematicMarketingShell`, one final cinematic reference stylesheet, and `ScrollVideoBackground`. Route content must not reintroduce the former public-site shell, extra headers, a footer, opaque SaaS cards, or competing route-level overrides.
+The first four routes use a shared cinematic implementation: `HomeV2` or `CinematicProductFunnel` supplies page data to `CinematicMarketingPage`; the shared page uses `CinematicMarketingShell`, one final cinematic reference stylesheet, and `ScrollVideoBackground`. The renderer samples a 24fps source into at most 120 cached frames (12fps) at up to 540px width, blends adjacent available frames, draws candidates to an offscreen staging canvas, retains the last good visible frame on failure, and stops its RAF loop when settled. Route content must not reintroduce extra headers, a footer, opaque SaaS cards, or competing route-level overrides.
 
-Contact, Docs, Privacy, and Terms retain their existing route-specific presentation and are outside this cinematic redesign. Keep their established implementation and legal meaning unchanged; use the repository's existing route checks to guard against regressions.
+Contact, Docs, Privacy, and Terms use the shared static utility shell: one transparent full-width nav, atmospheric poster treatment, Golos display headings, selective translucent glass, and no footer. Contact form handling and legal body copy are preserved. Docs uses current Evermind/Nevermind/Mastermind terminology and avoids animated video/canvas content behind reading material.
 
 ## Active work gates
 
 ### A. Renderer and motion
 
 - Reproduce slow/fast forward and reverse scroll in a real H.264-capable browser.
-- Prove cache and bootstrap handoffs do not expose blank frames; compare frames through blending and stop painting when settled.
+- Prove cache and bootstrap handoffs do not expose blank frames; stage candidate frames offscreen, retain the last good frame after draw failures, blend neighboring frames, and stop painting when settled.
+- Keep temporal sampling at 12fps (120 frames for the current 10.04-second media) within a roughly constant raw bitmap pixel budget by limiting each frame to 540px width.
 - Confirm the requested background demonstrably changes with scroll and chapter activation follows reading order.
 - Under reduced motion, freeze to one stable media/poster layer, disable scrubbing, preserve all content/links, and retain geometry.
 
 ### B. Visual and content QA
 
 - Check all four cinematic routes at desktop, laptop, tablet, and mobile widths; verify one H1, one main, no footer, no horizontal overflow, CTA wrapping, and consistent 100svh / 80vh / 100svh structure.
-- Do not redesign Contact, Docs, Privacy, or Terms in this release. Preserve their current route and legal behavior through existing regression tests.
+- Check Contact, Docs, Privacy, and Terms in the shared static utility shell; preserve contact submission behavior and all legal body meaning.
 - Review current copy against canonical Mind source and prevent stale product identities/claims in active pages and docs.
 - Capture and manually inspect viewport screenshots after motion settles, including top, transition, Section Two, page end, and reduced-motion desktop.
 - Prototype LiquidGlass locally and compare quality, CPU/frame cadence, media scrubbing, reduced motion, and initialization cost. Do not add a dependency or ship the effect without measured benefit.
 
-Initial LiquidGlass evaluation: a real Chrome local prototype using one dynamic scroll-canvas background and one cloned context panel ran at approximately 11–15fps (median frame interval 50–66.6ms) compared with the existing renderer's 16.7ms median cadence. The captured overlay also clipped/reflowed the cloned panel. Decision: do not add the dependency to the production scroll path; current CSS glass uses an 11% translucent white surface with 18px backdrop blur and preserves media throughput. This is a measured rejection for the current implementation, not a claim that the library cannot be used on static pages.
+LiquidGlass evaluation: a real Chrome local prototype using one dynamic scroll-canvas background and one cloned context panel ran at approximately 11–15fps (median frame interval 50–66.6ms) compared with the existing renderer's 16.7ms median cadence. The captured overlay also clipped/reflowed the cloned panel. Decision: do not add the dependency to the scroll path; retain the CSS glass fallback. A separate local Chrome 153 check of the revised cache recorded 89 visible canvas paints in 1.48s (59.99fps), longest paint gap 17.8ms, full 120-frame cache, and three down/up reversals without blanking or console errors.
 
 ### C. Validation and release
 
-Run the repository-supported env/doc integrity, TypeScript, ESLint, design lint, asset validation, production build, security, and relevant browser evidence. Review a diff-based safety pass; stage exact files only; commit and push the continuation branch without rewriting history. Integrate through the repository's normal `main` workflow. Verify workflow and deployment success, `/api/version` full SHA, all eight public routes, media and browser health. Then verify the authoritative worktree is clean and the protected checkout's branch, HEAD, status, and diff fingerprint are unchanged.
+Run the repository-supported env/doc integrity, TypeScript, ESLint, design lint, asset validation, production build, security, and relevant browser evidence. Review a diff-based safety pass; stage exact files only; commit and push the continuation branch without rewriting history. Integrate through the repository's normal `main` workflow. Verify workflow and deployment success, `/api/version` full SHA, all eight public routes, media and browser health. Then verify the authoritative worktree is clean and the protected checkout's branch, HEAD, status, and diff fingerprint are unchanged. Commit/push/deployment gates for this release are still pending.
 
 ## Known implementation/build debt
 
@@ -41,7 +42,7 @@ Run the repository-supported env/doc integrity, TypeScript, ESLint, design lint,
 
 ## Next performance gate
 
-Measure Lighthouse mobile LCP on `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, and `/contact`. The target remains ≤2.5 seconds. If missed, the next roadmap item is to bring canonical public-route LCP under 2.5 seconds without reducing the cinematic experience.
+The 2026-09-29 single-run local Lighthouse diagnostic measured LCP at `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, and `/contact` 4.21s. All exceed the ≤2.5s target; CLS was 0 and TBT 0ms. This is a diagnostic sample, not production field data. Next roadmap item: bring canonical public-route LCP under 2.5 seconds without reducing the cinematic experience; investigate the utility-route LCP increase separately.
 
 ## Completion evidence
 
diff --git a/docs/overview.md b/docs/overview.md
index 00f7125..497072b 100644
--- a/docs/overview.md
+++ b/docs/overview.md
@@ -7,8 +7,8 @@ This repository keeps active operating guidance separate from historical migrati
 Start here for present-day ProChat behavior:
 
 - `docs/repo-status.md` — current repository/product/runtime state
-- `docs/roadmap.md` — completed public roadmap plus explicitly deferred internal hardening
-- `docs/implementation-plan.md` — current implementation/hygiene acceptance criteria
+- `docs/roadmap.md` — current release status, completed work, performance follow-up, and explicitly deferred internal hardening
+- `docs/implementation-plan.md` — current renderer, utility-shell, validation, and release acceptance criteria
 - `docs/getting-started.md` — local setup and validation
 - `docs/development.md` — day-to-day contributor workflow
 - `docs/deployment.md` — deployment and production verification
@@ -36,7 +36,7 @@ The current public site surface is:
 - `/privacy`
 - `/terms`
 
-Evermind, Nevermind, and Mastermind are the current public product identities. ProChat Memory for QA is paused/historical. Home and the three product pages share the current cinematic navigation and scroll-driven media. Docs, Contact, and legal pages retain their existing page-specific presentation; they are outside this cinematic redesign scope.
+Evermind, Nevermind, and Mastermind are the current public product identities. ProChat Memory for QA is paused/historical. Home and the three product pages share the cinematic navigation and scroll-driven media. Docs, Contact, Privacy, and Terms use the same visual system through a static, reading-friendly utility shell; legal page body text is unchanged. Current publication status is recorded in `docs/repo-status.md` and `docs/roadmap.md`.
 
 ## Historical evidence
 
@@ -47,7 +47,7 @@ Historical plans and superseded implementation material belong in:
 - `docs/platform/**` when explicitly tied to completed historical PXF work
 - Git history
 
-Documents in those areas may mention retired ProChat OS, Kits, BuildFlow-as-product, Stripe, MailerLite, GitHub entitlement provisioning, older analytics, or older route strategies. They are evidence, not current product/runtime guidance.
+Documents in those areas may mention retired ProChat OS, Kits, BuildFlow-as-product, Stripe, MailerLite, GitHub entitlement provisioning, older analytics, or older route strategies. They are evidence, not current product/runtime guidance. Older `docs/homepage-*` specifications are pre-build design evidence; current implementation and QA authority is the shipped source, `docs/homepage-validation-plan.md`, `docs/implementation-plan.md`, and `docs/roadmap.md`.
 
 The final post-release hygiene pass additionally archives several retired system documents under:
 
diff --git a/docs/production-lifecycle.md b/docs/production-lifecycle.md
index 3c1e87a..006601a 100644
--- a/docs/production-lifecycle.md
+++ b/docs/production-lifecycle.md
@@ -56,15 +56,15 @@ A successful production closeout requires the `revision` returned by `/api/versi
 After deployment verify the eight canonical routes at desktop and mobile widths:
 
 - `/`
-- `/memory`
-- `/memory-qa`
-- `/workbench`
+- `/evermind`
+- `/nevermind`
+- `/mastermind`
 - `/docs`
 - `/contact`
 - `/privacy`
 - `/terms`
 
-Also verify intentional compatibility redirects and that retired product bodies remain unavailable.
+Also verify intentional compatibility redirects (`/workbench` and `/buildflow` to `/mastermind`) and that retired product bodies remain unavailable.
 
 ## Maintenance mode
 
diff --git a/docs/repo-status.md b/docs/repo-status.md
index b6aa4d7..9fe66eb 100644
--- a/docs/repo-status.md
+++ b/docs/repo-status.md
@@ -1,6 +1,6 @@
 # Repository status
 
-Status: current source has a locally validated cinematic implementation; production revision and route health are verified through the normal `main` workflow and `/api/version`.
+Status: the cinematic and utility-page implementation passes local validation. Starting production and `origin/main` revision is `c6731140191791e407a2fb6eb27ac96711883e32`; this release is not yet published or live-verified.
 
 Last reviewed: 2026-09-29.
 
@@ -12,8 +12,8 @@ Evermind and Nevermind are the connected Infinite Brain product system; Mastermi
 
 ## Runtime and visual architecture
 
-- `/`, `/evermind`, `/nevermind`, `/mastermind`: shared cinematic template, scroll-driven media, two chapters, 80vh intentional spacer, one navigation, and no footer.
-- `/contact`, `/docs`, `/privacy`, `/terms`: existing route-specific presentation, outside this visual redesign and checked for regressions only.
+- `/`, `/evermind`, `/nevermind`, `/mastermind`: shared cinematic template, scroll-driven media, two chapters, 80vh intentional spacer, one navigation, and no footer. The cache is sampled at 12fps (120 frames for the current 10.04s video) with each frame limited to 540px width; candidate rendering uses an offscreen staging canvas to preserve the last good visible frame.
+- `/contact`, `/docs`, `/privacy`, `/terms`: shared static utility shell with one transparent navigation and no footer. Contact submission remains functional; legal page body text is unchanged; Docs uses current product names and avoids animated background media.
 - Production integration: GitHub Actions `.github/workflows/main.yml` on `main`; production revision is checked through `/api/version`.
 
 ## Historical or deferred systems
@@ -22,7 +22,7 @@ Generated Docs/Nextra, retired checkout/licensing, Stripe runtime, MailerLite, G
 
 ## Validated release scope
 
-Renderer reverse-scroll/cache-handoff, responsive/reduced-motion and interaction evidence, measured LiquidGlass evaluation, current copy/documentation reconciliation, production build, browser/security checks, and manual desktop/mobile review are recorded for the cinematic release. Integrate through the normal GitHub `main` workflow and verify the deployed revision/routes before describing a publication as live. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
+Renderer continuity tests, responsive/reduced-motion and interaction evidence, measured LiquidGlass rejection, production build, 107 repository browser tests, nine focused quality tests, and manual desktop/mobile review are recorded for this release. Chrome 153 locally presented 89/90 tiny-scroll paints at 59.99fps (maximum observed gap 17.8ms), reached all 120 cache frames, and survived three full scroll reversals without blanking or console errors. Local single-run mobile LCP diagnostics: `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, `/contact` 4.21s; all exceed the 2.5s target. Production merge/deployment and remote source housekeeping remain pending. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
 
 ## Tracked performance/security notes
 
diff --git a/docs/roadmap.md b/docs/roadmap.md
index 97389e6..f88af56 100644
--- a/docs/roadmap.md
+++ b/docs/roadmap.md
@@ -1,6 +1,6 @@
 # ProChat website roadmap
 
-Status: cinematic implementation locally validated; use the normal `main` workflow and `/api/version` to determine the published revision. Performance and dependency hardening remain separate follow-up goals.
+Status: implementation and local release gates pass; current source is based on `main` at `c6731140191791e407a2fb6eb27ac96711883e32`. Production publication of this quality pass is pending the normal `main` release workflow and `/api/version` verification.
 
 Last reviewed: 2026-09-29.
 
@@ -8,7 +8,7 @@ Last reviewed: 2026-09-29.
 
 The active public routes are `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, `/contact`, `/privacy`, and `/terms`. Evermind and Nevermind are the connected Infinite Brain product system; Mastermind is the separate free planning/orchestration product. ProChat Memory for QA is paused/historical. Mind remains canonical for product and company strategy.
 
-The first four routes share a two-section cinematic page, one full-width navigation, one scroll-driven background renderer, one 80vh spacer, and no footer. Contact, Docs, Privacy, and Terms are regression-checked but not redesigned in this release; preserve their current route presentation and legal meaning.
+The first four routes share a two-section cinematic page, one full-width navigation, one scroll-driven background renderer, one 80vh spacer, and no footer. Contact and Docs now use the same visual language through a static, reading-friendly utility shell; Privacy and Terms use that same shell while their legal body text remains unchanged. All four utility pages have one navigation and no footer.
 
 ## Current release goal — ProChat cinematic/site quality pass
 
@@ -16,17 +16,18 @@ The release source of truth is `feature/prochat-cinematic-finalization-2026-09`;
 
 In scope:
 
-1. Fix and verify smooth reverse/forward scroll, cache/bootstrap frame blending, chapter pacing, and CTA wrapping on the four cinematic product routes.
+1. Fix and verify reverse/forward continuity and smooth frame presentation, chapter pacing, and CTA wrapping on the four cinematic product routes.
 2. Keep those four routes visually consistent, including Golos display headings, one navigation, translucent capability glass, correct 100svh / 80vh / 100svh geometry, and zero footers.
-3. Keep Contact, Docs, Privacy, and Terms outside the visual redesign; retain existing routes and verify their established CI contracts.
-4. Verify prefers-reduced-motion, 1440/1280/tablet/mobile layouts, route content, menu interactions, browser errors, and manually inspected cinematic screenshots.
-5. Evaluate LiquidGlass with a measured local prototype; do not ship it unless it materially improves the glass treatment without harming media scroll performance, accessibility, or initial rendering. Initial 1440×900 Chrome prototype result: not accepted for this scroll-driven page—the one-panel dynamic-canvas run measured about 11–15fps (50–66.6ms median RAF interval) versus 16.7ms baseline cadence, and the prototype overlay did not preserve the card composition. Keep the current 11% CSS liquid-glass treatment.
-6. Audit current public copy against Mind, correct active website guidance, validate, review, commit, push, merge through the normal `main` workflow, then verify the live revision and routes.
-7. Preserve the protected checkout and unrelated repositories/worktrees. After confirmed production success only, reconcile Mastermind source continuity, indexes, and safely removable branches/worktrees; do not delete unique or unverified work.
+3. Bring Contact and Docs into the shared ProChat visual system with restrained/static backgrounds; migrate only shared chrome on Privacy and Terms, preserving legal meaning.
+4. Verify prefers-reduced-motion, 1440/1280/1024/768/430/390/360 layouts, route content, menu and Contact interactions, browser errors, and manually inspected screenshots.
+5. LiquidGlass was tested in a local 1440×900 Chrome prototype on the hero context card and capability panel. It was not adopted: dynamic-canvas performance was about 11–15fps (50–66.6ms median RAF interval) versus 16.7ms baseline cadence, and the overlay clipped/reflowed the composition. Retain the CSS glass fallback; do not add the dependency to the scroll path.
+6. Current local Chrome 153 renderer check reached the 120-frame cache and presented 89/90 tiny-scroll paints at 59.99fps, with a longest observed gap of 17.8ms; three full down/up reversals left the canvas painted and produced no console errors. This is a local desktop diagnostic, not a cross-device field guarantee.
+7. Update active implementation docs, validate, review, commit, push, merge through the normal `main` workflow, then verify the live revision and routes. Publication is not complete until that verification passes.
+8. Preserve the protected checkout and unrelated repositories/worktrees. After confirmed production success only, reconcile Mastermind source continuity, indexes, and safely removable branches/worktrees; do not delete unique or unverified work.
 
 ## Performance follow-up
 
-The canonical mobile LCP target is at most 2.5 seconds. A single-run local production-artifact diagnostic on 2026-09-29 (Lighthouse 12.4, Moto G Power 2022 simulation, 4× CPU, slow 4G; diagnostic mode) measured LCP as follows: `/` 2.83s, `/evermind` 3.19s, `/nevermind` 2.79s, `/mastermind` 2.80s, `/docs` 2.72s, `/contact` 2.99s. All six exceeded target; CLS was 0 and TBT was 0ms. These are local lab measurements, not production field data, and one run per route is directional rather than a stable baseline. The next website goal is: **Bring canonical public-route LCP under 2.5 seconds without reducing cinematic fidelity.** Do not trade away the requested scroll-driven visuals to satisfy the metric.
+The canonical mobile LCP target is at most 2.5 seconds. A single-run local production-artifact diagnostic on 2026-09-29 (Lighthouse 12.4, Moto G Power 2022 simulation, 4× CPU, slow 4G; diagnostic mode) measured: `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, `/contact` 4.21s. All six exceeded target; CLS was 0 and TBT was 0ms. Utility-page LCP needs follow-up investigation. These are local lab measurements, not production field data; one run per route is directional rather than a stable baseline. The next website goal is: **Bring canonical public-route LCP under 2.5 seconds without reducing cinematic fidelity.** Do not trade away the requested scroll-driven visuals to satisfy the metric.
 
 ## Tracked build hardening
 
diff --git a/docs/token-architecture.md b/docs/token-architecture.md
index eaee8ae..1e68f3b 100644
--- a/docs/token-architecture.md
+++ b/docs/token-architecture.md
@@ -1,10 +1,12 @@
 # ProChat Token Architecture
 
-**Status:** canonical repository-local token governance
+**Status:** historical PXF-010 token proposal; not the active route-design authority
 **Canonical design source:** Mind `wiki/organisations/prochat/brand/global-design-foundation.md`, translated by `brand-spec.md`, operationalized by `DESIGN.md`
 **Date:** 2026-07-29
 **Phase:** PXF-010 — Design-system governance and launch hardening
 
+> This document records an earlier foundation/public/`--pm-*` token proposal. Do not treat its statements about current token completeness, route consumption, or required migration steps as verified current implementation. The active public route contract is documented in `docs/overview.md`, `docs/implementation-plan.md`, and `docs/homepage-validation-plan.md`. The core cinematic routes are styled by `src/components/cinematic/cinematic-reference-final.css`; static utility routes use `src/components/cinematic/cinematic-utility.css`. Verify source before changing shared tokens.
+
 ## Token layers
 
 The system uses three hierarchical token layers:
diff --git a/docs/website-copy-blueprint.md b/docs/website-copy-blueprint.md
index 312011e..a169527 100644
--- a/docs/website-copy-blueprint.md
+++ b/docs/website-copy-blueprint.md
@@ -1,9 +1,11 @@
 # ProChat website copy blueprint
 
-**Status:** active website copy guidance  
+**Status:** superseded pre-cinematic copy plan; retained as historical design evidence
 **Owner:** Steve Westhoek  
 **Scope:** homepage, product pages, niche page, contact/waitlist copy, navigation, footer, metadata, and social copy
 
+> This file predates the current Evermind/Nevermind/Mastermind product and website decisions. Do not use its copy examples or hierarchy as current guidance. The current public copy is in the route source, with product truth in Mind; current implementation status and checks are in `docs/roadmap.md`, `docs/implementation-plan.md`, and `docs/homepage-validation-plan.md`.
+
 ## Authority
 
 Mind is canonical for ProChat philosophy, product hierarchy, naming, positioning, business stage, growth policy, legal-policy direction, and cross-product roadmap.
diff --git a/src/app/(marketing)/components/layout/PublicLegalPage.tsx b/src/app/(marketing)/components/layout/PublicLegalPage.tsx
index 256a469..8cf4d55 100644
--- a/src/app/(marketing)/components/layout/PublicLegalPage.tsx
+++ b/src/app/(marketing)/components/layout/PublicLegalPage.tsx
@@ -1,4 +1,5 @@
 import type { ReactNode } from 'react'
+import { CinematicMarketingShell } from '@/components/CinematicMarketingShell'
 
 import '@/assets/styles/prochat-public-chrome.css'
 import '@/assets/styles/prochat-public-bodies.css'
@@ -19,7 +20,12 @@ export function PublicLegalPage({
   summary,
 }: PublicLegalPageProps) {
   return (
-    <div className="pc-body-page public-legal-page__main pc-legal-ledger" data-body-family="legal" data-legal-kind={kind}>
+    <CinematicMarketingShell
+      contentOwnsMain
+      utilityTheme
+      cta={{ label: 'Explore Evermind', href: '/evermind', trackingCta: 'explore_evermind' }}
+    >
+    <main id="main-content" className="pc-body-page public-legal-page__main pc-legal-ledger" data-body-family="legal" data-legal-kind={kind}>
       <section className="pc-legal-ledger__masthead" aria-labelledby={`legal-${kind}-title`}>
         <div className="pc-legal-ledger__masthead-inner">
           <div>
@@ -38,6 +44,7 @@ export function PublicLegalPage({
       </section>
 
       {children}
-    </div>
+    </main>
+    </CinematicMarketingShell>
   )
 }
diff --git a/src/app/(marketing)/contact/ContactPageClient.tsx b/src/app/(marketing)/contact/ContactPageClient.tsx
index 0f03999..a5ca144 100644
--- a/src/app/(marketing)/contact/ContactPageClient.tsx
+++ b/src/app/(marketing)/contact/ContactPageClient.tsx
@@ -7,7 +7,6 @@ import { trackEvent } from '@/utils/analytics'
 import { CinematicMarketingShell } from '@/components/CinematicMarketingShell'
 
 import ContactPageMarkup from './ContactPageMarkup'
-import { CinematicMediaPage } from '@/components/cinematic/CinematicMediaPage'
 
 type ContactFieldName =
   | 'name'
@@ -34,16 +33,13 @@ const FORM_FIELDS: ContactFieldName[] = [
 
 const CONTACT_SUBMIT_IDLE_HTML = `
   <span class="pc-action-label">
-    <span class="text-current md:hidden">SEND MY BRIEF</span>
-    <span class="hidden text-current md:inline">SEND MY BRIEF</span>
-    <span aria-hidden="true" class="hidden opacity-50 md:inline"> - </span>
-    <span class="hidden opacity-50 md:inline">FOCUSED MEMORY CONTEXT</span>
+    <span class="text-current">SEND MESSAGE</span>
   </span>
 `
 
 const CONTACT_SUBMIT_SUBMITTING_HTML = `
   <span class="pc-action-label">
-    <span class="text-current">SENDING BRIEF...</span>
+    <span class="text-current">SENDING MESSAGE...</span>
   </span>
 `
 
@@ -297,12 +293,11 @@ export default function ContactPageClient({ initialTopic }: ContactPageClientPro
   return (
     <CinematicMarketingShell
       contentOwnsMain
+      utilityTheme
       cta={{ label: 'Start with Evermind', href: '/evermind', trackingCta: 'start_evermind' }}
     >
       <main id="main-content" ref={rootRef} className="contact-page-root">
-        <CinematicMediaPage className="cm-contact-experience" chapterCount={2} lite>
-          <ContactPageMarkup initialTopic={initialTopic} />
-        </CinematicMediaPage>
+        <ContactPageMarkup initialTopic={initialTopic} />
       </main>
     </CinematicMarketingShell>
   )
diff --git a/src/app/(marketing)/contact/ContactPageMarkup.tsx b/src/app/(marketing)/contact/ContactPageMarkup.tsx
index 7d3e5bb..3048359 100644
--- a/src/app/(marketing)/contact/ContactPageMarkup.tsx
+++ b/src/app/(marketing)/contact/ContactPageMarkup.tsx
@@ -10,17 +10,14 @@ const TOPIC_CONTEXT = {
   Evermind: {
     label: 'Evermind enquiry',
     helper: 'Describe the memory workflow or local-first question you want to explore.',
-    signals: ['Memory use case', 'Human ownership', 'Clear next step'],
   },
   Nevermind: {
     label: 'Nevermind enquiry',
     helper: 'Describe the context problem or AI-tool workflow you want to make more continuous.',
-    signals: ['Working context', 'Tool choice', 'Clear next step'],
   },
   Mastermind: {
     label: 'Mastermind enquiry',
     helper: 'Describe the intention, project context, or bounded execution workflow you want to clarify.',
-    signals: ['Project context', 'Execution boundary', 'Clear next step'],
   },
 } as const
 
@@ -30,54 +27,35 @@ export default function ContactPageMarkup({
   const topicContext = TOPIC_CONTEXT[initialTopic]
 
   return (
-    <div className="cm-contact-page" data-body-family="contact">
-      <section className="cm-chapter cm-contact-hero" data-cinematic-chapter="0" aria-labelledby="contact-hero-title">
-        <div className="cm-chapter__inner">
-          <p className="cm-chapter__eyebrow">Contact / direct signal</p>
-          <h1 id="contact-hero-title">Send the context.<br /><span>Get the clearest next step.</span></h1>
-          <p className="cm-lede">Tell me what you are working on, where the friction is, and what outcome you need. The goal is one useful reply with the right next step.</p>
-          <div className="cm-contact-signals">{topicContext.signals.map((signal, index) => <span key={signal}><b>0{index + 1}</b>{signal}</span>)}</div>
-          <a className="cm-actions__primary cm-contact-hero__action" href="#contact-form-card">Send the brief <span aria-hidden="true">→</span></a>
+    <div className="cm-utility-contact" data-body-family="contact">
+      <section className="cm-utility-contact__intro" aria-labelledby="contact-hero-title">
+        <div className="pc-body-kicker"><span aria-hidden="true" />Contact ProChat</div>
+        <h1 id="contact-hero-title">Start a conversation.</h1>
+        <p>Tell us what you are working on and what would help. Choose Evermind, Nevermind, or Mastermind if you have a product-specific question, and we will reply with a useful next step.</p>
+        <div className="cm-utility-contact__topic" role="group" aria-label="Selected contact topic">
+          <span>ABOUT</span>
+          <strong>{topicContext.label}</strong>
+          <small>{topicContext.helper}</small>
         </div>
       </section>
 
-      <section className="cm-chapter cm-contact-intake" data-cinematic-chapter="1" aria-labelledby="contact-intake-title">
-        <div className="cm-chapter__inner contact-intake-grid">
-          <aside className="contact-intake-context">
-            <div className="pc-body-kicker"><span aria-hidden="true" />Request context</div>
-            <h2 id="contact-intake-title">One brief is enough to start.</h2>
-            <p>{topicContext.helper}</p>
-
-            <div className="contact-topic-console" role="group" aria-label="Selected contact topic">
-              <span>SELECTED ROUTE</span>
-              <strong>{topicContext.label}</strong>
-              <small>Replies are reviewed manually.</small>
-            </div>
-
-            <ol className="contact-response-rail" aria-label="What happens next">
-              <li><span>01</span><strong>You send context</strong><p>Enough detail to understand the real problem.</p></li>
-              <li><span>02</span><strong>I review the fit</strong><p>No automated qualification funnel.</p></li>
-              <li><span>03</span><strong>You get a next step</strong><p>A direct answer, question, or recommended path.</p></li>
-            </ol>
-          </aside>
-
-          <div className="contact-form-panel cm-glass-panel" id="contact-form-card">
-            <div className="contact-form-panel__header">
+      <section className="cm-utility-contact__form contact-form-panel" aria-labelledby="contact-form-title">
+          <div className="contact-form-panel__header">
               <div>
-                <span>INTAKE / 01</span>
-                <h2>Send a message</h2>
+                <span>MESSAGE / 01</span>
+                <h2 id="contact-form-title">Send a message</h2>
               </div>
               <span className="contact-form-status-dot">OPEN</span>
-            </div>
-            <p>{topicContext.helper}</p>
-
-            <form
-              data-contact-form=""
-              className="contact-memory-form"
-              noValidate
-              method="post"
-              action="/api/contact"
-            >
+          </div>
+          <p>Required fields are marked. A few clear sentences are enough.</p>
+
+          <form
+            data-contact-form=""
+            className="contact-memory-form"
+            noValidate
+            method="post"
+            action="/api/contact"
+          >
               <div className="sr-only" aria-hidden="true">
                 <label htmlFor="contact-honeypot">Leave this empty</label>
                 <input
@@ -101,7 +79,7 @@ export default function ContactPageMarkup({
                     id="contact-name"
                     name="name"
                     className="contact-field"
-                    placeholder="John Doe"
+                    placeholder="Your name"
                     autoComplete="name"
                     aria-required="true"
                     required
@@ -117,7 +95,7 @@ export default function ContactPageMarkup({
                     id="contact-email"
                     name="email"
                     className="contact-field"
-                    placeholder="john@company.com"
+                    placeholder="you@example.com"
                     type="email"
                     autoComplete="email"
                     aria-required="true"
@@ -167,9 +145,7 @@ export default function ContactPageMarkup({
                 />
                 <span aria-hidden="true">→</span>
               </button>
-            </form>
-          </div>
-        </div>
+          </form>
       </section>
     </div>
   )
diff --git a/src/app/docs/layout.tsx b/src/app/docs/layout.tsx
index 27609fa..970d7b9 100644
--- a/src/app/docs/layout.tsx
+++ b/src/app/docs/layout.tsx
@@ -1,5 +1,14 @@
 import type { ReactNode } from 'react'
+import { CinematicMarketingShell } from '@/components/CinematicMarketingShell'
 
 export default function DocsLayout({ children }: { children: ReactNode }) {
-  return children
+  return (
+    <CinematicMarketingShell
+      contentOwnsMain
+      utilityTheme
+      cta={{ label: 'Explore Evermind', href: '/evermind', trackingCta: 'explore_evermind' }}
+    >
+      <div data-utility-main="">{children}</div>
+    </CinematicMarketingShell>
+  )
 }
diff --git a/src/app/docs/page.tsx b/src/app/docs/page.tsx
index 0929b75..9ccb327 100644
--- a/src/app/docs/page.tsx
+++ b/src/app/docs/page.tsx
@@ -1,5 +1,4 @@
 import StructuredData from '@/components/StructuredData'
-import PublicEditorialHero from '@/components/public/PublicEditorialHero'
 import { getSEOTags } from '@/libs/seo'
 import { getDocsSchemas } from '@/libs/structured-data'
 
@@ -28,8 +27,8 @@ export async function generateMetadata() {
     },
     socialImage: {
       line1: 'ProChat Documentation',
-      line2: 'Memory + context + execution',
-      subtitle: 'Three products. One connected direction.',
+      line2: 'Evermind + Nevermind + Mastermind',
+      subtitle: 'Product documentation for AI work.',
     },
     canonicalUrlRelative: '/docs',
   })
@@ -79,24 +78,14 @@ const paths = [
 
 export default function DocsIndexPage() {
   return (
-    <main className="pc-body-page pc-docs-hub" data-body-family="docs">
+    <main className="pc-body-page pc-docs-hub cm-utility-docs" data-body-family="docs">
       <StructuredData id="schema-docs" data={getDocsSchemas()} />
 
-      <PublicEditorialHero
-        variant="docs"
-        eyebrow="Documentation / product family"
-        title={
-          <>
-            Three products.<br />
-            <strong>One connected direction.</strong>
-          </>
-        }
-        description="Explore the product family from human-owned memory to working context and controlled execution. Each product has a distinct role, and the pages keep those boundaries explicit."
-        primaryAction={{ href: '#repository-docs', label: 'Choose a product' }}
-        signals={['Evermind', 'Nevermind', 'Mastermind']}
-        visualTitle="DOCS / PRODUCT FAMILY"
-        visualCaption="REMEMBER → CONTEXTUALIZE → DIRECT"
-      />
+      <header className="cm-utility-docs__hero">
+        <div className="pc-body-kicker"><span aria-hidden="true" />ProChat / Documentation</div>
+        <h1>Product documentation for AI work.</h1>
+        <p>{DOCS_DESCRIPTION} Start with the product that matches the work, then use its documentation to understand setup, boundaries, and next steps.</p>
+      </header>
 
       <section className="pc-docs-hub__paths" id="repository-docs" aria-labelledby="repository-docs-title">
         <div className="pc-docs-hub__intro">
diff --git a/src/components/CinematicMarketingShell.tsx b/src/components/CinematicMarketingShell.tsx
index 8c985b5..da37b18 100644
--- a/src/components/CinematicMarketingShell.tsx
+++ b/src/components/CinematicMarketingShell.tsx
@@ -9,6 +9,7 @@ import { SocialIcon } from '@/components/ui/social-icons'
 import { trackEvent } from '@/utils/analytics'
 import './cinematic-marketing-shell.css'
 import './cinematic/cinematic-reference-final.css'
+import './cinematic/cinematic-utility.css'
 
 const NAV_ITEMS = [
   { href: '/evermind', label: 'Evermind' },
@@ -197,18 +198,20 @@ export function CinematicMarketingShell({
   cta,
   contentOwnsMain = false,
   cinematicTheme = false,
+  utilityTheme = false,
 }: {
   children: React.ReactNode
   cta?: CinematicMarketingCta
   contentOwnsMain?: boolean
   cinematicTheme?: boolean
+  utilityTheme?: boolean
 }) {
   return (
-    <div className={`cm-shell ${cinematicTheme ? 'cm-shell--core' : ''}`.trim()}>
+    <div className={`cm-shell ${cinematicTheme ? 'cm-shell--core' : ''} ${utilityTheme ? 'cm-shell--utility' : ''}`.trim()}>
       <div className="cm-shell__backdrop" aria-hidden="true" />
       <CinematicMarketingNav cta={cta} />
       {contentOwnsMain ? children : <main id="main-content" className="cm-shell__main">{children}</main>}
-      {!cinematicTheme ? <CinematicMarketingFooter /> : null}
+      {!cinematicTheme && !utilityTheme ? <CinematicMarketingFooter /> : null}
     </div>
   )
 }
diff --git a/src/components/cinematic/cinematic-utility.css b/src/components/cinematic/cinematic-utility.css
new file mode 100644
index 0000000..e0bc9b0
--- /dev/null
+++ b/src/components/cinematic/cinematic-utility.css
@@ -0,0 +1,334 @@
+/* Quiet utility-page expression of the shared ProChat cinematic identity. */
+.cm-shell.cm-shell--utility {
+  --cm-utility-white: 255 255 255;
+  --cm-utility-ink: 15 34 48;
+  position: relative;
+  min-height: 100svh;
+  overflow-x: clip;
+  background: oklch(15% 0.012 242);
+  color: rgb(var(--cm-utility-white));
+  font-family: var(--font-sans), system-ui, sans-serif;
+}
+
+.cm-shell.cm-shell--utility > .cm-shell__backdrop {
+  position: fixed;
+  z-index: 0;
+  inset: 0;
+  background-image:
+    linear-gradient(180deg, rgb(8 16 24 / 0.56), rgb(8 16 24 / 0.74)),
+    url('/funnels/cinematic-bootstrap.jpg');
+  background-repeat: no-repeat;
+  background-size: cover, 600% 400%;
+  background-position: center, 0 0;
+}
+
+.cm-shell.cm-shell--utility > .cm-nav-shell {
+  position: fixed;
+  z-index: 50;
+  inset: 0 0 auto;
+  width: 100%;
+  margin: 0;
+  padding: 0;
+}
+
+.cm-shell.cm-shell--utility .cm-nav {
+  width: 100%;
+  min-height: 4.875rem;
+  padding: 0 3rem;
+  border: 0;
+  border-bottom: 1px solid rgb(255 255 255 / 0.15);
+  border-radius: 0;
+  background: rgb(20 31 41 / 0.12);
+  box-shadow: none;
+  backdrop-filter: blur(8px);
+}
+
+.cm-shell.cm-shell--utility .cm-nav__brand,
+.cm-shell.cm-shell--utility .cm-nav__links a,
+.cm-shell.cm-shell--utility .cm-nav__contact,
+.cm-shell.cm-shell--utility .cm-nav__links a:hover,
+.cm-shell.cm-shell--utility .cm-nav__links a[aria-current='page'],
+.cm-shell.cm-shell--utility .cm-nav__contact:hover,
+.cm-shell.cm-shell--utility .cm-nav__menu-button {
+  color: rgb(var(--cm-utility-white));
+}
+
+.cm-shell.cm-shell--utility .cm-nav__cta {
+  min-height: 2.75rem;
+  border-color: rgb(255 255 255 / 0.32);
+  background: rgb(255 255 255 / 0.95);
+  color: rgb(var(--cm-utility-ink));
+  white-space: nowrap;
+}
+
+.cm-shell.cm-shell--utility .cm-nav__menu-button {
+  border-color: rgb(255 255 255 / 0.28);
+  background: rgb(255 255 255 / 0.1);
+}
+
+.cm-shell.cm-shell--utility .cm-nav__mobile-panel {
+  border-color: rgb(255 255 255 / 0.22);
+  background: rgb(29 43 54 / 0.94);
+  box-shadow: 0 1.5rem 4rem rgb(0 0 0 / 0.24);
+  backdrop-filter: blur(18px);
+}
+
+.cm-shell.cm-shell--utility .cm-nav__mobile-panel a {
+  color: rgb(255 255 255 / 0.9);
+}
+
+.cm-shell.cm-shell--utility > main,
+.cm-shell.cm-shell--utility > [data-utility-main] {
+  position: relative;
+  z-index: 5;
+  padding-top: 4.875rem;
+}
+
+.cm-shell.cm-shell--utility :is(h1, h2, h3) {
+  font-family: var(--font-brand), 'Golos Text', sans-serif;
+  text-wrap: balance;
+}
+
+.cm-shell.cm-shell--utility .cm-utility-contact,
+.cm-shell.cm-shell--utility .cm-utility-docs,
+.cm-shell.cm-shell--utility .pc-legal-ledger {
+  --cm-line: rgb(255 255 255 / 0.18);
+  --cm-muted: rgb(245 248 251 / 0.78);
+  color: rgb(255 255 255 / 0.96);
+}
+
+.cm-shell.cm-shell--utility .cm-utility-contact {
+  width: min(calc(100% - 3rem), 82rem);
+  display: grid;
+  grid-template-columns: minmax(0, 0.8fr) minmax(0, 1.2fr);
+  align-items: start;
+  gap: clamp(2rem, 7vw, 6rem);
+  margin-inline: auto;
+  padding-block: clamp(3.5rem, 8vw, 7rem);
+}
+
+.cm-shell.cm-shell--utility .cm-utility-contact__intro { padding-top: 1.2rem; }
+.cm-shell.cm-shell--utility .cm-utility-contact__intro h1 {
+  max-width: 10ch;
+  margin: 1rem 0 0;
+  font-size: clamp(3rem, 6.6vw, 6.8rem);
+  font-weight: 500;
+  letter-spacing: -0.055em;
+  line-height: 0.98;
+}
+.cm-shell.cm-shell--utility .cm-utility-contact__intro > p {
+  max-width: 34rem;
+  margin: 1.35rem 0 0;
+  color: var(--cm-muted);
+  font-size: clamp(1rem, 1.3vw, 1.13rem);
+  line-height: 1.7;
+}
+.cm-shell.cm-shell--utility .cm-utility-contact__topic {
+  display: grid;
+  gap: 0.35rem;
+  margin-top: 1.8rem;
+  padding: 1rem 1.1rem;
+  border: 1px solid rgb(255 255 255 / 0.18);
+  border-radius: 0.85rem;
+  background: rgb(255 255 255 / 0.1);
+  backdrop-filter: blur(12px);
+}
+.cm-shell.cm-shell--utility .cm-utility-contact__topic span,
+.cm-shell.cm-shell--utility .cm-utility-contact__topic small {
+  color: rgb(255 255 255 / 0.7);
+  font-size: 0.76rem;
+}
+.cm-shell.cm-shell--utility .cm-utility-contact__topic strong { font-weight: 600; }
+.cm-shell.cm-shell--utility .cm-utility-contact__form {
+  padding: clamp(1.25rem, 3vw, 2.2rem);
+  border: 1px solid rgb(255 255 255 / 0.22);
+  border-radius: 1.2rem;
+  background: rgb(255 255 255 / 0.12);
+  box-shadow: inset 0 1px 0 rgb(255 255 255 / 0.17), 0 1.5rem 4rem rgb(9 25 38 / 0.2);
+  backdrop-filter: blur(18px);
+}
+.cm-shell.cm-shell--utility .cm-utility-contact__form > p,
+.cm-shell.cm-shell--utility .contact-helper-text,
+.cm-shell.cm-shell--utility .contact-form-panel .text-muted-foreground {
+  color: rgb(255 255 255 / 0.76);
+}
+.cm-shell.cm-shell--utility .contact-form-panel__header { display: flex; align-items: start; justify-content: space-between; gap: 1rem; }
+.cm-shell.cm-shell--utility .contact-form-panel__header h2 { margin: 0.4rem 0 0; font-size: clamp(1.8rem, 3vw, 2.5rem); }
+.cm-shell.cm-shell--utility .contact-memory-form { display: grid; gap: 1.1rem; margin-top: 1.6rem; }
+.cm-shell.cm-shell--utility .contact-memory-grid { display: grid; grid-template-columns: repeat(2, minmax(0, 1fr)); gap: 1rem; }
+.cm-shell.cm-shell--utility .contact-memory-field { min-width: 0; display: grid; gap: 0.45rem; }
+.cm-shell.cm-shell--utility .contact-field,
+.cm-shell.cm-shell--utility .contact-textarea { min-height: 2.9rem; padding: 0.72rem 0.8rem; }
+.cm-shell.cm-shell--utility .contact-textarea { min-height: 8rem; resize: vertical; }
+.cm-shell.cm-shell--utility .contact-form-panel .pc-body-button {
+  width: 100%;
+  min-height: 2.9rem;
+  display: inline-flex;
+  align-items: center;
+  justify-content: center;
+  gap: 0.65rem;
+  padding: 0.7rem 1rem;
+  border: 1px solid rgb(255 255 255 / 0.35);
+  font: 600 0.82rem/1.2 var(--font-sans), system-ui, sans-serif;
+  cursor: pointer;
+}
+.cm-shell.cm-shell--utility .contact-form-panel .pc-body-button:disabled { cursor: wait; opacity: 0.65; }
+.cm-shell.cm-shell--utility .contact-field-error { margin: 0; color: rgb(255 210 210); font-size: 0.78rem; }
+.cm-shell.cm-shell--utility .contact-status { margin: 0; color: white; }
+.cm-shell.cm-shell--utility .contact-form-panel__header h2,
+.cm-shell.cm-shell--utility .contact-field-label { color: white; }
+.cm-shell.cm-shell--utility .contact-form-panel__header > div > span,
+.cm-shell.cm-shell--utility .contact-topic-console span,
+.cm-shell.cm-shell--utility .contact-form-panel__header > .contact-form-status-dot {
+  color: rgb(255 255 255 / 0.72);
+}
+.cm-shell.cm-shell--utility .contact-topic-console {
+  border: 1px solid rgb(255 255 255 / 0.18);
+  border-radius: 0.75rem;
+  background: rgb(255 255 255 / 0.08);
+}
+.cm-shell.cm-shell--utility .contact-field,
+.cm-shell.cm-shell--utility .contact-textarea {
+  border-color: rgb(255 255 255 / 0.25);
+  border-radius: 0.65rem;
+  background: rgb(13 26 37 / 0.48);
+  color: white;
+}
+.cm-shell.cm-shell--utility .contact-field::placeholder,
+.cm-shell.cm-shell--utility .contact-textarea::placeholder { color: rgb(255 255 255 / 0.56); }
+.cm-shell.cm-shell--utility .contact-field:focus-visible,
+.cm-shell.cm-shell--utility .contact-textarea:focus-visible {
+  border-color: rgb(255 255 255 / 0.75);
+  outline: 2px solid rgb(255 255 255 / 0.3);
+  outline-offset: 2px;
+}
+.cm-shell.cm-shell--utility .contact-form-panel .pc-body-button {
+  border-radius: 999px;
+  background: rgb(255 255 255 / 0.96);
+  color: rgb(var(--cm-utility-ink));
+  white-space: nowrap;
+}
+
+.cm-shell.cm-shell--utility .cm-utility-docs {
+  width: min(calc(100% - 3rem), 82rem);
+  margin-inline: auto;
+  padding-block: clamp(3rem, 7vw, 6rem);
+}
+.cm-shell.cm-shell--utility .cm-utility-docs__hero {
+  max-width: 54rem;
+  margin-bottom: clamp(2.5rem, 6vw, 4.5rem);
+}
+.cm-shell.cm-shell--utility .cm-utility-docs__hero h1 {
+  margin: 1rem 0 0;
+  font-size: clamp(3rem, 6.8vw, 6.6rem);
+  font-weight: 500;
+  letter-spacing: -0.055em;
+  line-height: 0.98;
+}
+.cm-shell.cm-shell--utility .cm-utility-docs__hero p {
+  max-width: 44rem;
+  margin: 1.25rem 0 0;
+  color: var(--cm-muted);
+  font-size: clamp(1rem, 1.25vw, 1.12rem);
+  line-height: 1.7;
+}
+.cm-shell.cm-shell--utility .pc-docs-hub__intro h2 {
+  max-width: 18ch;
+  margin: 0.8rem 0 0;
+  font-size: clamp(2rem, 4vw, 3.4rem);
+  font-weight: 500;
+  letter-spacing: -0.04em;
+}
+.cm-shell.cm-shell--utility .pc-docs-hub__paths {
+  display: grid;
+  grid-template-columns: minmax(15rem, 0.65fr) minmax(0, 1.35fr);
+  align-items: start;
+  gap: clamp(2rem, 5vw, 4rem);
+}
+.cm-shell.cm-shell--utility .pc-docs-hub__intro > p,
+.cm-shell.cm-shell--utility .pc-docs-hub__card > p,
+.cm-shell.cm-shell--utility .pc-docs-hub__boundary > p {
+  color: var(--cm-muted);
+  line-height: 1.7;
+}
+.cm-shell.cm-shell--utility .pc-docs-hub__grid {
+  display: grid;
+  grid-template-columns: repeat(3, minmax(0, 1fr));
+  gap: 1rem;
+  margin-top: 1.8rem;
+}
+.cm-shell.cm-shell--utility .pc-docs-hub__card {
+  min-width: 0;
+  display: flex;
+  flex-direction: column;
+  padding: clamp(1.2rem, 2vw, 1.8rem);
+  border: 1px solid rgb(255 255 255 / 0.2);
+  border-radius: 1rem;
+  background: rgb(255 255 255 / 0.1);
+  box-shadow: inset 0 1px 0 rgb(255 255 255 / 0.13);
+  backdrop-filter: blur(14px);
+}
+.cm-shell.cm-shell--utility .pc-docs-hub__card h3 { margin: 0.2rem 0 0; color: white; font-size: 1.45rem; }
+.cm-shell.cm-shell--utility .pc-docs-hub__card small,
+.cm-shell.cm-shell--utility .pc-docs-hub__card header > span { color: rgb(255 255 255 / 0.72); }
+.cm-shell.cm-shell--utility .pc-docs-hub__signals { display: flex; flex-wrap: wrap; gap: 0.45rem; margin: 1.2rem 0; padding: 0; list-style: none; }
+.cm-shell.cm-shell--utility .pc-docs-hub__signals li { padding: 0.35rem 0.55rem; border: 1px solid rgb(255 255 255 / 0.18); border-radius: 999px; color: rgb(255 255 255 / 0.82); font-size: 0.72rem; }
+.cm-shell.cm-shell--utility .pc-docs-hub__links { display: grid; gap: 0.65rem; }
+.cm-shell.cm-shell--utility .pc-docs-hub__links a { display: flex; justify-content: space-between; gap: 0.5rem; color: white; font-size: 0.86rem; font-weight: 600; text-decoration: none; }
+.cm-shell.cm-shell--utility .pc-docs-hub__links a:hover { text-decoration: underline; text-underline-offset: 0.2em; }
+.cm-shell.cm-shell--utility .pc-docs-hub__boundary {
+  display: grid;
+  grid-template-columns: minmax(0, 1fr) minmax(14rem, 0.75fr);
+  gap: 2rem;
+  margin-top: 3rem;
+  padding-top: 1.5rem;
+  border-top: 1px solid rgb(255 255 255 / 0.2);
+}
+.cm-shell.cm-shell--utility .pc-docs-hub__boundary h2 { max-width: 22ch; margin: 0.8rem 0 0; font-size: 1.2rem; }
+
+.cm-shell.cm-shell--utility .pc-legal-ledger { padding-bottom: 4rem; }
+.cm-shell.cm-shell--utility .pc-legal-ledger__masthead,
+.cm-shell.cm-shell--utility .pc-legal-ledger__body { background: transparent; }
+.cm-shell.cm-shell--utility .pc-legal-ledger__masthead { padding-block: clamp(2rem, 5vw, 4rem); }
+.cm-shell.cm-shell--utility .pc-legal-ledger__masthead-inner,
+.cm-shell.cm-shell--utility .pc-legal-ledger__body > div { width: min(calc(100% - 3rem), 78rem); }
+.cm-shell.cm-shell--utility .pc-legal-ledger__masthead h1,
+.cm-shell.cm-shell--utility .pc-legal-ledger__article h2,
+.cm-shell.cm-shell--utility .pc-legal-ledger__article h3 { color: white; }
+.cm-shell.cm-shell--utility .pc-legal-ledger__masthead p,
+.cm-shell.cm-shell--utility .pc-legal-ledger__article { color: rgb(255 255 255 / 0.84); }
+.cm-shell.cm-shell--utility .pc-legal-ledger__register,
+.cm-shell.cm-shell--utility .pc-legal-ledger__article aside,
+.cm-shell.cm-shell--utility .pc-legal-ledger__article blockquote {
+  border-color: rgb(255 255 255 / 0.2);
+  background: rgb(255 255 255 / 0.09);
+  color: rgb(255 255 255 / 0.84);
+}
+.cm-shell.cm-shell--utility .pc-legal-ledger a { color: white; }
+
+@media (max-width: 1100px) {
+  .cm-shell.cm-shell--utility .pc-docs-hub__paths { grid-template-columns: 1fr; }
+}
+
+@media (max-width: 900px) {
+  .cm-shell.cm-shell--utility .cm-nav { padding-inline: 2rem; }
+  .cm-shell.cm-shell--utility .cm-utility-contact { grid-template-columns: 1fr; gap: 2rem; }
+  .cm-shell.cm-shell--utility .pc-docs-hub__paths { grid-template-columns: 1fr; }
+  .cm-shell.cm-shell--utility .pc-docs-hub__grid { grid-template-columns: 1fr; }
+}
+
+@media (max-width: 620px) {
+  .cm-shell.cm-shell--utility .cm-nav { padding-inline: 1rem; }
+  .cm-shell.cm-shell--utility .cm-utility-contact,
+  .cm-shell.cm-shell--utility .cm-utility-docs { width: min(calc(100% - 2rem), 82rem); }
+  .cm-shell.cm-shell--utility .contact-memory-grid { grid-template-columns: 1fr; }
+  .cm-shell.cm-shell--utility .pc-docs-hub__boundary { grid-template-columns: 1fr; gap: 0.75rem; }
+  .cm-shell.cm-shell--utility .pc-legal-ledger__masthead-inner,
+  .cm-shell.cm-shell--utility .pc-legal-ledger__body > div { width: calc(100% - 2rem); }
+}
+
+@media (prefers-reduced-motion: reduce) {
+  .cm-shell.cm-shell--utility *,
+  .cm-shell.cm-shell--utility *::before,
+  .cm-shell.cm-shell--utility *::after { scroll-behavior: auto !important; }
+}
diff --git a/src/components/funnels/ScrollVideoBackground.tsx b/src/components/funnels/ScrollVideoBackground.tsx
index a8ee63e..9253d71 100644
--- a/src/components/funnels/ScrollVideoBackground.tsx
+++ b/src/components/funnels/ScrollVideoBackground.tsx
@@ -10,8 +10,9 @@ const BOOTSTRAP_FRAME_COUNT = 24
 const BOOTSTRAP_COLUMNS = 6
 const BOOTSTRAP_ROWS = 4
 const FIRST_FRAME_TIME_SECONDS = 0.04
-const MAX_CACHED_FRAMES = 72
-const MAX_CACHED_FRAME_WIDTH = 720
+const MAX_CACHED_FRAMES = 120
+const CACHE_READY_FRAME_COUNT = 48
+const MAX_CACHED_FRAME_WIDTH = 540
 
 type ScrollVideoBackgroundProps = {
   containerRef?: RefObject<HTMLElement | null>
@@ -44,6 +45,7 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
   const targetRef = useRef(0)
   const smoothedRef = useRef(0)
   const onProgressRef = useRef(onProgress)
+  const wakeRendererRef = useRef<() => void>(() => {})
   const [bootstrapReady, setBootstrapReady] = useState(false)
   const [videoReady, setVideoReady] = useState(false)
   const [cacheReady, setCacheReady] = useState(false)
@@ -69,6 +71,7 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
         : getProgress(containerRef?.current ?? null)
       targetRef.current = result.progress
       onProgressRef.current?.(result.progress, result.isActive)
+      wakeRendererRef.current()
     }
 
     updateTarget()
@@ -173,7 +176,7 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
       })
       const duration = source.duration
       if (!Number.isFinite(duration) || duration <= 0) return
-      const count = Math.min(MAX_CACHED_FRAMES, Math.max(24, Math.round(duration * 10)))
+      const count = Math.min(MAX_CACHED_FRAMES, Math.max(24, Math.round(duration * 12)))
       const width = Math.min(MAX_CACHED_FRAME_WIDTH, source.videoWidth || MAX_CACHED_FRAME_WIDTH)
       const height = Math.max(1, Math.round(width * ((source.videoHeight || 540) / (source.videoWidth || 960))))
       const temp = document.createElement('canvas')
@@ -200,7 +203,7 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
         framesRef.current[index]?.close()
         framesRef.current[index] = frame
         loadedCount += 1
-        if (loadedCount === BOOTSTRAP_FRAME_COUNT) setCacheReady(true)
+        if (loadedCount === CACHE_READY_FRAME_COUNT) setCacheReady(true)
       }
     }
 
@@ -221,6 +224,8 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
     let raf = 0
     let previousTime = 0
     let forceDraw = true
+    const stagingCanvas = document.createElement('canvas')
+    const stagingContext = stagingCanvas.getContext('2d')
     const drawCover = (
       ctx: CanvasRenderingContext2D,
       source: CanvasImageSource,
@@ -263,23 +268,23 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
       smoothedRef.current += (targetRef.current - smoothedRef.current) * (1 - Math.exp(-delta / 88))
       const moved = Math.abs(smoothedRef.current - oldProgress) > 0.00001
       const canvas = canvasRef.current
-      const canDrawCanvas = canvas && bootstrapReady
+      const canDrawCanvas = canvas && bootstrapReady && stagingContext
 
       if (canDrawCanvas) {
         const dpr = Math.min(window.devicePixelRatio || 1, 1.5)
         const width = Math.round(window.innerWidth * dpr)
         const height = Math.round(window.innerHeight * dpr)
-        let resized = false
-        if (canvas.width !== width || canvas.height !== height) {
-          canvas.width = width
-          canvas.height = height
-          resized = true
+        const resized = stagingCanvas.width !== width || stagingCanvas.height !== height
+        if (resized) {
+          stagingCanvas.width = width
+          stagingCanvas.height = height
         }
-        const ctx = canvas.getContext('2d')
 
-        if (ctx && (forceDraw || resized || moved)) {
+        if (forceDraw || resized || moved) {
           let drawn = false
+          const ctx = stagingContext
           try {
+            ctx.clearRect(0, 0, width, height)
             if (cacheReady && frameCountRef.current > 0) {
               const frames = framesRef.current
               const position = smoothedRef.current * (frameCountRef.current - 1)
@@ -331,28 +336,45 @@ export function ScrollVideoBackground({ containerRef, className = '', onProgress
               if (upperIndex !== lowerIndex && blend > 0) drawBootstrap(upperIndex, false, blend)
               drawn = true
             }
-          } catch {
-            // Keep the last successfully painted canvas frame if a source is transiently unavailable.
-          }
+          } catch { /* A failed candidate never touches the visible canvas. */ }
 
           if (drawn) {
-            forceDraw = false
-            if (!canvasPainted) setCanvasPainted(true)
+            try {
+              if (canvas.width !== width || canvas.height !== height) {
+                canvas.width = width
+                canvas.height = height
+              }
+              const visibleContext = canvas.getContext('2d')
+              if (visibleContext) {
+                visibleContext.drawImage(stagingCanvas, 0, 0, width, height)
+                forceDraw = false
+                if (!canvasPainted) setCanvasPainted(true)
+              }
+            } catch { /* Keep the last successfully presented frame. */ }
           }
         }
       }
 
-      if (!prefersReducedMotion || !canvasPainted && bootstrapReady) {
+      raf = 0
+      if (Math.abs(targetRef.current - smoothedRef.current) > 0.00001) {
         raf = requestAnimationFrame(draw)
       }
     }
 
-    raf = requestAnimationFrame(draw)
-    return () => cancelAnimationFrame(raf)
+    const wake = () => {
+      if (raf) return
+      raf = requestAnimationFrame(draw)
+    }
+    wakeRendererRef.current = wake
+    wake()
+    return () => {
+      wakeRendererRef.current = () => {}
+      if (raf) cancelAnimationFrame(raf)
+    }
   }, [bootstrapReady, cacheReady, canvasPainted, prefersReducedMotion])
 
   const canvasVisible = canvasPainted && !prefersReducedMotion
   const videoHidden = canvasPainted || prefersReducedMotion
   const renderState = cacheReady ? 'cache' : canvasPainted ? 'bootstrap' : 'poster'
-  return <><link rel="preload" as="image" href={BOOTSTRAP_SPRITE_URL} /><div className={`cm-video ${className}`.trim()} aria-hidden="true" data-render-state={renderState}><div className={`cm-video__poster ${canvasVisible ? 'is-hidden' : ''}`} /><video ref={videoRef} className={`cm-video__element ${videoHidden ? 'is-hidden' : videoReady ? 'is-visible' : ''}`} src={VIDEO_URL} muted playsInline preload="metadata" /><canvas ref={canvasRef} className={`cm-video__canvas ${canvasVisible ? 'is-visible' : ''}`} data-frame-cache-ready={cacheReady ? 'true' : 'false'} /><div className="cm-video__veil" /></div></>
+  return <><link rel="preload" as="image" href={BOOTSTRAP_SPRITE_URL} /><div className={`cm-video ${className}`.trim()} aria-hidden="true" data-render-state={renderState}><div className={`cm-video__poster ${canvasVisible ? 'is-hidden' : ''}`} /><video ref={videoRef} className={`cm-video__element ${videoHidden ? 'is-hidden' : videoReady ? 'is-visible' : ''}`} src={VIDEO_URL} muted playsInline preload="metadata" /><canvas ref={canvasRef} className={`cm-video__canvas ${canvasVisible ? 'is-visible' : ''}`} data-frame-cache-ready={cacheReady ? 'true' : 'false'} data-frame-count={cacheReady ? frameCountRef.current : '0'} data-canvas-painted={canvasPainted ? 'true' : 'false'} /><div className="cm-video__veil" /></div></>
 }
diff --git a/src/helpers/public-route-design.ts b/src/helpers/public-route-design.ts
index 1003dc9..94f2a14 100644
--- a/src/helpers/public-route-design.ts
+++ b/src/helpers/public-route-design.ts
@@ -20,9 +20,13 @@ export function getUnifiedPublicRouteConfig(pathname: string): PublicRouteDesign
   const path = normalize(pathname)
 
   if (path === '/') return { variant: 'home', contentOwnsShell: true }
-  if (path === '/docs') return { variant: 'docs', contentOwnsMain: true }
+  if (path === '/docs' || path.startsWith('/docs/')) {
+    return { variant: 'docs', contentOwnsShell: true }
+  }
   if (path === '/contact') return { variant: 'contact', contentOwnsShell: true }
-  if (path === '/privacy' || path === '/terms') return { variant: 'legal' }
+  if (path === '/privacy' || path === '/terms') {
+    return { variant: 'legal', contentOwnsShell: true }
+  }
 
   return null
 }
diff --git a/tests/evidence/canonical-chrome-proof.spec.ts b/tests/evidence/canonical-chrome-proof.spec.ts
index 17bba4a..6692e2d 100644
--- a/tests/evidence/canonical-chrome-proof.spec.ts
+++ b/tests/evidence/canonical-chrome-proof.spec.ts
@@ -16,7 +16,9 @@ const STANDARD_PUBLIC_ROUTES = [
 ] as const
 
 const CINEMATIC_PRODUCT_ROUTES = ['/evermind', '/nevermind', '/mastermind'] as const
-const CINEMATIC_MARKETING_ROUTES = new Set<string>(['/', '/contact', ...CINEMATIC_PRODUCT_ROUTES])
+const CINEMATIC_MARKETING_ROUTES = new Set<string>([
+  '/', '/contact', '/docs', '/privacy', '/terms', ...CINEMATIC_PRODUCT_ROUTES,
+])
 
 const DESKTOP = { name: 'desktop', width: 1440, height: 1000 } as const
 const MOBILE = { name: 'mobile', width: 390, height: 900 } as const
@@ -69,7 +71,7 @@ test.describe('canonical public chrome — structure and first-paint invariants'
 
         // Core cinematic pages intentionally end after Section Two; other routes retain one footer.
         const footer = publicFooter(page, route)
-        const expectedFooterCount = CORE_CINEMATIC_ROUTES.has(route) ? 0 : 1
+        const expectedFooterCount = 0
         expect(await footer.count(), `${route} public footer count at ${viewport.name}`).toBe(expectedFooterCount)
         if (expectedFooterCount) await expect(footer).toBeVisible()
 
@@ -275,8 +277,8 @@ test.describe('docs page — repository hub with canonical shell', () => {
       waitUntil: 'domcontentloaded',
     })
 
-    await expect(page.locator('nav.pm-navbar')).toBeVisible()
-    await expect(page.locator('footer.pc-footer')).toBeVisible()
+    await expect(page.locator('nav.cm-nav')).toBeVisible()
+    await expect(page.locator('footer')).toHaveCount(0)
 
     const docsHub = page.locator('main.pc-docs-hub')
     await expect(docsHub).toBeVisible()
@@ -298,10 +300,10 @@ test.describe('docs page — repository hub with canonical shell', () => {
       waitUntil: 'domcontentloaded',
     })
 
-    await expect(page.locator('nav.pm-navbar')).toBeVisible()
+    await expect(page.locator('nav.cm-nav')).toBeVisible()
     await expect(page.locator('main.pc-docs-hub')).toBeVisible()
     await expect(page.locator('.pc-docs-hub__card')).toHaveCount(3)
-    await expect(page.locator('footer.pc-footer')).toBeVisible()
+    await expect(page.locator('footer')).toHaveCount(0)
 
     const layout = await page.evaluate(() => ({
       documentWidth: document.documentElement.scrollWidth,
@@ -345,17 +347,17 @@ test.describe('contact page — canonical copy and layout', () => {
     })
 
     await expect(page.locator('nav.cm-nav')).toBeVisible()
-    await expect(page.locator('footer.cm-footer')).toBeVisible()
-    await expect(page.locator('.cm-contact-page')).toBeVisible()
-    await expect(page.locator('.contact-intake-grid')).toBeVisible()
+    await expect(page.locator('footer')).toHaveCount(0)
+    await expect(page.locator('.cm-utility-contact')).toBeVisible()
+    await expect(page.locator('.cm-utility-contact__intro')).toBeVisible()
     await expect(page.locator('.contact-form-panel')).toBeVisible()
-    await expect(page.getByText('Send the context', { exact: false })).toBeVisible()
-    await expect(page.getByText('One brief is enough to start.', { exact: false })).toBeVisible()
+    await expect(page.getByRole('heading', { name: 'Start a conversation.' })).toBeVisible()
+    await expect(page.getByRole('heading', { name: 'Send a message' })).toBeVisible()
 
     const layout = await page.evaluate(() => ({
       documentWidth: document.documentElement.scrollWidth,
       viewportWidth: window.innerWidth,
-      intakeWidth: document.querySelector<HTMLElement>('.contact-intake-grid')?.getBoundingClientRect().width ?? 0,
+      intakeWidth: document.querySelector<HTMLElement>('.cm-utility-contact')?.getBoundingClientRect().width ?? 0,
       formHeight: document.querySelector<HTMLElement>('.contact-form-panel')?.getBoundingClientRect().height ?? 0,
     }))
 
@@ -372,7 +374,7 @@ test.describe('contact page — canonical copy and layout', () => {
 
     await expect(page.locator('nav.cm-nav')).toBeVisible()
     await expect(page.locator('.contact-form-panel')).toBeVisible()
-    await expect(page.locator('footer.cm-footer')).toBeVisible()
+    await expect(page.locator('footer')).toHaveCount(0)
 
     const layout = await page.evaluate(() => ({
       documentWidth: document.documentElement.scrollWidth,
@@ -425,8 +427,8 @@ test.describe('client navigation — chrome integrity across route changes', ()
     ).toBe(1)
     expect(
       await publicFooter(page, '/docs').count(),
-      'exactly one footer after navigating to /docs',
-    ).toBe(1)
+      'no footer after navigating to /docs',
+    ).toBe(0)
     // No skip control visible after navigation
     await expect(
       page.getByRole('link', { name: /^skip to content$/i }),
@@ -441,8 +443,8 @@ test.describe('client navigation — chrome integrity across route changes', ()
     ).toBe(1)
     expect(
       await publicFooter(page, '/contact').count(),
-      'exactly one footer after navigating to /contact',
-    ).toBe(1)
+      'no footer after navigating to /contact',
+    ).toBe(0)
 
     // Navigate back to homepage
     await page.goto(new URL('/', baseUrl).toString(), { waitUntil: 'domcontentloaded' })
diff --git a/tests/evidence/canonical-route-smoke.spec.ts b/tests/evidence/canonical-route-smoke.spec.ts
index a6701c9..96198c1 100644
--- a/tests/evidence/canonical-route-smoke.spec.ts
+++ b/tests/evidence/canonical-route-smoke.spec.ts
@@ -7,7 +7,7 @@ if (!baseUrl) {
 }
 
 const routes = ['/', '/evermind', '/nevermind', '/mastermind', '/docs', '/contact', '/privacy', '/terms'] as const
-const cinematicRoutes = new Set<string>(['/', '/evermind', '/nevermind', '/mastermind', '/contact'])
+const cinematicRoutes = new Set<string>(['/', '/evermind', '/nevermind', '/mastermind'])
 const coreCinematicRoutes = new Set<string>(['/', '/evermind', '/nevermind', '/mastermind'])
 
 const viewports = [
@@ -50,23 +50,20 @@ test.describe('canonical route smoke evidence', () => {
         await expect(main).toBeVisible()
 
         if (cinematicRoutes.has(route)) {
-          if (route === '/' || route === '/contact') {
+          if (route === '/') {
             await expect(page.locator('[data-cinematic-experience]'), `${route} is missing its cinematic experience`).toHaveCount(1)
           } else {
             await expect(page.locator('.cm-canonical-main'), `${route} is missing its cinematic root`).toHaveCount(1)
             await expect(page.locator('.cm-canonical-main')).toBeVisible()
           }
           await expect(page.locator('nav.cm-nav'), `${route} is missing its cinematic navigation`).toHaveCount(1)
-          await expect(page.locator('footer.cm-footer'), `${route} footer contract`).toHaveCount(coreCinematicRoutes.has(route) ? 0 : 1)
+          await expect(page.locator('footer.cm-footer'), `${route} footer contract`).toHaveCount(0)
         } else {
-          const navigation = page.locator('nav.pm-navbar')
+          const navigation = page.locator('nav[aria-label="Primary navigation"]')
+          await expect(page.locator('.cm-shell--utility'), `${route} is missing the shared utility shell`).toHaveCount(1)
           await expect(navigation, `${route} is missing the canonical public navigation`).toHaveCount(1)
           await expect(navigation).toBeVisible()
-
-          const footer = page.locator('footer.pc-footer')
-          await expect(footer, `${route} is missing the canonical public footer`).toHaveCount(1)
-          await expect(footer).toBeVisible()
-          await expect(footer.getByRole('link', { name: 'ProChat home' })).toBeVisible()
+          await expect(page.locator('footer'), `${route} utility pages do not render a legacy footer`).toHaveCount(0)
         }
 
         // A non-empty primary heading must be present — proves the page rendered.
@@ -99,15 +96,15 @@ test.describe('contact page visual closeout', () => {
     await page.setViewportSize({ width: 1440, height: 1000 })
     await page.goto(new URL('/contact', baseUrl).toString(), { waitUntil: 'networkidle' })
 
-    await expect(page.locator('.cm-contact-page')).toBeVisible()
-    await expect(page.locator('.contact-intake-grid')).toBeVisible()
+    await expect(page.locator('.cm-utility-contact')).toBeVisible()
+    await expect(page.locator('.cm-utility-contact__intro')).toBeVisible()
     await expect(page.locator('.contact-form-panel')).toBeVisible()
-    await expect(page.getByText('Send the context', { exact: false })).toBeVisible()
-    await expect(page.getByText('One brief is enough to start.', { exact: false })).toBeVisible()
+    await expect(page.getByRole('heading', { name: 'Start a conversation.' })).toBeVisible()
+    await expect(page.getByRole('heading', { name: 'Send a message' })).toBeVisible()
 
     const evidence = await page.evaluate(() => {
       const shell = document.querySelector('.cm-shell') as HTMLElement | null
-      const intake = document.querySelector('.contact-intake-grid') as HTMLElement | null
+      const intake = document.querySelector('.cm-utility-contact') as HTMLElement | null
       const panel = document.querySelector('.contact-form-panel') as HTMLElement | null
       const shellStyle = shell ? getComputedStyle(shell) : null
       return {
@@ -133,9 +130,9 @@ test.describe('contact page visual closeout', () => {
     await page.goto(new URL('/contact', baseUrl).toString(), { waitUntil: 'networkidle' })
 
     await expect(page.locator('nav.cm-nav')).toBeVisible()
-    await expect(page.locator('.contact-intake-grid')).toBeVisible()
+    await expect(page.locator('.cm-utility-contact')).toBeVisible()
     await expect(page.locator('.contact-form-panel')).toBeVisible()
-    await expect(page.locator('footer.cm-footer')).toBeVisible()
+    await expect(page.locator('footer')).toHaveCount(0)
 
     const layout = await page.evaluate(() => ({
       documentWidth: document.documentElement.scrollWidth,
diff --git a/tests/evidence/cinematic-final-quality.spec.ts b/tests/evidence/cinematic-final-quality.spec.ts
index c8d771e..e7ce49c 100644
--- a/tests/evidence/cinematic-final-quality.spec.ts
+++ b/tests/evidence/cinematic-final-quality.spec.ts
@@ -71,6 +71,20 @@ async function sampleRenderedCanvas(page: Page) {
   })
 }
 
+async function renderedCanvasChecksum(page: Page) {
+  return page.locator('.cm-video__canvas').evaluate(element => {
+    const canvas = element as HTMLCanvasElement
+    const sample = document.createElement('canvas')
+    sample.width = 24
+    sample.height = 14
+    const context = sample.getContext('2d')
+    if (!context) throw new Error('Canvas 2D context is unavailable')
+    context.drawImage(canvas, 0, 0, sample.width, sample.height)
+    return [...context.getImageData(0, 0, sample.width, sample.height).data]
+      .reduce((sum, value, index) => (sum + value * (index + 1)) % 1_000_000_007, 0)
+  })
+}
+
 test.describe('ProChat final public-site quality', () => {
   test('core routes keep every cinematic background painted through repeated scroll reversals', async ({ page }) => {
     test.setTimeout(120_000)
@@ -81,16 +95,125 @@ test.describe('ProChat final public-site quality', () => {
       const response = await page.goto(routeUrl(route), { waitUntil: 'domcontentloaded' })
       expect(response?.status(), `${route} response`).toBe(200)
       await expect(page.locator('.cm-video__canvas.is-visible')).toBeVisible()
+      await expect(page.locator('.cm-video__canvas')).toHaveAttribute('data-canvas-painted', 'true')
       await expect.poll(() => page.locator('.cm-video__canvas').evaluate(canvas => getComputedStyle(canvas).opacity)).toBe('1')
 
+      const canDecodeVideo = await page.evaluate(() => Boolean(
+        document.createElement('video').canPlayType('video/mp4; codecs="avc1.640028"'),
+      ))
+      if (canDecodeVideo) {
+        await expect(page.locator('.cm-video__canvas')).toHaveAttribute('data-frame-cache-ready', 'true', { timeout: 45_000 })
+        await expect(page.locator('.cm-video__canvas')).toHaveAttribute('data-frame-count', '120')
+      }
+
       const sampled = await sampleRenderedCanvas(page)
       expect(sampled.samples, `${route} sampled frame count`).toBeGreaterThan(200)
       expect(sampled.visibleCanvasSamples, `${route} visible canvas samples`).toBeGreaterThan(200)
       expect(sampled.uncoveredSamples, `${route} uncovered background samples`).toBe(0)
       expect(sampled.states.every(state => state === 'bootstrap' || state === 'cache'), `${route} media state stays painted`).toBe(true)
+      if (canDecodeVideo) expect(sampled.states, `${route} observed cache handoff`).toContain('cache')
+    }
+  })
+
+  test('a failed candidate frame leaves the last good frame visible on all core routes', async ({ page }) => {
+    test.setTimeout(120_000)
+    await page.setViewportSize({ width: 1440, height: 900 })
+    await page.emulateMedia({ reducedMotion: 'no-preference' })
+
+    for (const route of CORE_ROUTES) {
+      await page.goto(routeUrl(route), { waitUntil: 'domcontentloaded' })
+      await expect(page.locator('.cm-video__canvas')).toHaveAttribute('data-canvas-painted', 'true')
+      const beforeFailure = await renderedCanvasChecksum(page)
+
+      await page.evaluate(() => {
+        const visibleCanvas = document.querySelector<HTMLCanvasElement>('.cm-video__canvas')
+        if (!visibleCanvas) throw new Error('Visible cinematic canvas is missing')
+        const prototype = CanvasRenderingContext2D.prototype
+        const original = prototype.drawImage
+        const testWindow = window as Window & { __restoreCinematicDrawImage?: typeof original }
+        testWindow.__restoreCinematicDrawImage = original
+        prototype.drawImage = new Proxy(original, {
+          apply(target, thisArg: CanvasRenderingContext2D, args: Parameters<typeof original>) {
+            if (
+              thisArg.canvas !== visibleCanvas &&
+              thisArg.canvas.width === visibleCanvas.width &&
+              thisArg.canvas.height === visibleCanvas.height
+            ) {
+              const injectedWindow = window as Window & { __cinematicCandidateFailure?: boolean }
+              injectedWindow.__cinematicCandidateFailure = true
+              throw new Error('Injected one-frame decode failure')
+            }
+            return Reflect.apply(target, thisArg, args)
+          },
+        })
+        window.scrollBy({ top: 260, behavior: 'instant' })
+      })
+
+      await expect.poll(() => page.evaluate(() => Boolean(
+        (window as Window & { __cinematicCandidateFailure?: boolean }).__cinematicCandidateFailure,
+      ))).toBe(true)
+      expect(await renderedCanvasChecksum(page), `${route} preserves last good pixels after a failed candidate`).toBe(beforeFailure)
+      await page.waitForTimeout(180)
+      expect(await renderedCanvasChecksum(page), `${route} continues to show a painted frame`).toBeGreaterThan(0)
+
+      await page.evaluate(() => {
+        const original = (window as Window & { __restoreCinematicDrawImage?: CanvasRenderingContext2D['drawImage'] }).__restoreCinematicDrawImage
+        if (original) CanvasRenderingContext2D.prototype.drawImage = original
+        delete (window as Window & { __cinematicCandidateFailure?: boolean }).__cinematicCandidateFailure
+        delete (window as Window & { __restoreCinematicDrawImage?: CanvasRenderingContext2D['drawImage'] }).__restoreCinematicDrawImage
+      })
     }
   })
 
+  test('cached renderer presents tiny scroll changes above 30 frames per second', async ({ page }) => {
+    await page.setViewportSize({ width: 1440, height: 900 })
+    await page.emulateMedia({ reducedMotion: 'no-preference' })
+    await page.goto(routeUrl('/'), { waitUntil: 'domcontentloaded' })
+    const canDecodeVideo = await page.evaluate(() => Boolean(
+      document.createElement('video').canPlayType('video/mp4; codecs="avc1.640028"'),
+    ))
+    test.skip(!canDecodeVideo, 'This browser has no supported H.264 decoder for the cached-video path.')
+    await expect(page.locator('.cm-video__canvas')).toHaveAttribute('data-frame-cache-ready', 'true', { timeout: 45_000 })
+
+    const cadence = await page.evaluate(async () => {
+      const canvas = document.querySelector<HTMLCanvasElement>('.cm-video__canvas')
+      if (!canvas) throw new Error('Cinematic canvas is missing')
+      const prototype = CanvasRenderingContext2D.prototype
+      const original = prototype.drawImage
+      const paintTimes: number[] = []
+      prototype.drawImage = new Proxy(original, {
+        apply(target, thisArg: CanvasRenderingContext2D, args: Parameters<typeof original>) {
+          if (thisArg.canvas === canvas) paintTimes.push(performance.now())
+          return Reflect.apply(target, thisArg, args)
+        },
+      })
+
+      try {
+        window.scrollTo({ top: 0, behavior: 'instant' })
+        await new Promise<void>(resolve => requestAnimationFrame(() => resolve()))
+        for (let index = 0; index < 90; index += 1) {
+          window.scrollBy({ top: index % 2 ? -1.5 : 2, behavior: 'instant' })
+          await new Promise<void>(resolve => requestAnimationFrame(() => resolve()))
+        }
+      } finally {
+        prototype.drawImage = original
+      }
+
+      const intervals = paintTimes.slice(1).map((time, index) => time - paintTimes[index])
+      const duration = paintTimes.length > 1 ? paintTimes.at(-1)! - paintTimes[0] : 0
+      return {
+        samples: paintTimes.length,
+        duration,
+        framesPerSecond: duration > 0 ? (paintTimes.length - 1) * 1000 / duration : 0,
+        longestGap: Math.max(0, ...intervals),
+      }
+    })
+
+    expect(cadence.samples, 'tiny scroll movements repaint the visible canvas').toBeGreaterThan(45)
+    expect(cadence.framesPerSecond, `observed paint cadence: ${JSON.stringify(cadence)}`).toBeGreaterThanOrEqual(30)
+    expect(cadence.longestGap, `no long gaps during fine scroll: ${JSON.stringify(cadence)}`).toBeLessThan(100)
+  })
+
   test('four cinematic pages share one clean header and CTA labels stay on a single line at all target widths', async ({ page }) => {
     test.setTimeout(180_000)
     await page.emulateMedia({ reducedMotion: 'no-preference' })
@@ -104,7 +227,7 @@ test.describe('ProChat final public-site quality', () => {
         const evidence = await page.evaluate(() => {
           const heading = document.querySelector<HTMLElement>('main h1')
           const ctas = Array.from(document.querySelectorAll<HTMLElement>(
-            '.cm-actions a, .cm-nav__cta',
+            '.cm-actions a, .cm-nav__cta, .cm-template-context-card .cm-inline-link',
           )).filter(element => element.getBoundingClientRect().width > 0).map(element => {
             const textNode = Array.from(element.childNodes).find(node => node.nodeType === Node.TEXT_NODE && node.textContent?.trim())
             let lineCount = 0
@@ -209,3 +332,105 @@ test.describe('ProChat final public-site quality', () => {
     }), { timeout: 10_000 }).not.toBe(firstFrame)
   })
 })
+
+test.describe('ProChat utility-page cinematic shell', () => {
+  test('Contact, Docs, Privacy, and Terms share the quiet utility shell without legacy chrome or animation', async ({ page }) => {
+    await page.setViewportSize({ width: 1440, height: 900 })
+    await page.emulateMedia({ reducedMotion: 'no-preference' })
+
+    for (const route of ['/contact', '/docs', '/privacy', '/terms']) {
+      const response = await page.goto(routeUrl(route), { waitUntil: 'domcontentloaded' })
+      expect(response?.status(), `${route} response`).toBe(200)
+      const evidence = await page.evaluate(() => {
+        const nav = document.querySelector<HTMLElement>('nav[aria-label="Primary navigation"]')
+        const h1 = document.querySelector<HTMLElement>('main h1')
+        const main = document.querySelector<HTMLElement>('main')
+        return {
+          shell: document.querySelectorAll('.cm-shell--utility').length,
+          navs: document.querySelectorAll('nav[aria-label="Primary navigation"]').length,
+          navShells: document.querySelectorAll('.cm-nav-shell').length,
+          mains: document.querySelectorAll('main').length,
+          footers: document.querySelectorAll('footer').length,
+          legacyNavs: document.querySelectorAll('.pm-navbar,.cpf-nav').length,
+          navRadius: nav ? getComputedStyle(nav).borderRadius : 'missing',
+          headingFont: h1 ? getComputedStyle(h1).fontFamily : '',
+          mediaRenderer: document.querySelectorAll('.cm-video,.cm-video__canvas,video').length,
+          overflow: document.documentElement.scrollWidth > window.innerWidth,
+          mainText: main?.innerText ?? '',
+        }
+      })
+
+      expect(evidence.shell, `${route} utility shell`).toBe(1)
+      expect(evidence.navs, `${route} primary navigation`).toBe(1)
+      expect(evidence.navShells, `${route} header shell count`).toBe(1)
+      expect(evidence.mains, `${route} main landmark`).toBe(1)
+      expect(evidence.footers, `${route} no footer`).toBe(0)
+      expect(evidence.legacyNavs, `${route} no legacy capsule`).toBe(0)
+      expect(evidence.navRadius, `${route} full-width navigation`).toBe('0px')
+      expect(evidence.headingFont, `${route} Golos display`).toMatch(/Golos[_ ]Text/i)
+      expect(evidence.mediaRenderer, `${route} static utility backdrop`).toBe(0)
+      expect(evidence.overflow, `${route} no horizontal overflow`).toBe(false)
+      if (route === '/contact' || route === '/docs') {
+        expect(evidence.mainText, `${route} current public product names`).not.toMatch(/ProChat Memory|Memory for QA|Workbench/i)
+      }
+    }
+  })
+
+  test('Contact keeps an accessible form and handles a successful submission', async ({ page }) => {
+    await page.setViewportSize({ width: 390, height: 844 })
+    await page.emulateMedia({ reducedMotion: 'reduce' })
+    await page.route('**/api/contact', async route => route.fulfill({
+      status: 200,
+      contentType: 'application/json',
+      body: JSON.stringify({ message: 'Thanks, your message has been sent.' }),
+    }))
+    await page.goto(routeUrl('/contact'), { waitUntil: 'networkidle' })
+    await page.getByLabel('Name').fill('Taylor Example')
+    await page.getByLabel('Email').fill('taylor@example.com')
+    await page.getByRole('textbox', { name: 'Message' }).fill('I would like to learn about Evermind.')
+    await page.getByRole('button', { name: /send message/i }).click()
+    await expect(page.locator('[data-contact-status]')).toContainText('Thanks, your message has been sent.')
+    await expect(page.getByRole('navigation', { name: 'Primary navigation' })).toBeVisible()
+  })
+
+  test('Contact, Docs, Privacy, and Terms remain usable across the responsive matrix', async ({ page }) => {
+    for (const width of VIEWPORTS) {
+      await page.setViewportSize({ width, height: width < 600 ? 844 : 900 })
+      for (const route of ['/contact', '/docs', '/privacy', '/terms']) {
+        const response = await page.goto(routeUrl(route), { waitUntil: 'domcontentloaded' })
+        expect(response?.status(), `${route} at ${width}px`).toBe(200)
+
+        const evidence = await page.evaluate(() => {
+          const heading = document.querySelector<HTMLElement>('main h1')
+          const cta = document.querySelector<HTMLElement>('.cm-nav__cta')
+          return {
+            shellCount: document.querySelectorAll('.cm-shell--utility').length,
+            navCount: document.querySelectorAll('nav[aria-label="Primary navigation"]').length,
+            mainCount: document.querySelectorAll('main').length,
+            footerCount: document.querySelectorAll('footer').length,
+            overflow: document.documentElement.scrollWidth > window.innerWidth,
+            headingVisible: Boolean(heading && heading.getBoundingClientRect().width > 0 && heading.innerText.trim()),
+            ctaOverflow: Boolean(cta && cta.scrollWidth > cta.clientWidth + 1),
+          }
+        })
+
+        expect(evidence.shellCount, `${route} utility shell at ${width}px`).toBe(1)
+        expect(evidence.navCount, `${route} nav at ${width}px`).toBe(1)
+        expect(evidence.mainCount, `${route} main at ${width}px`).toBe(1)
+        expect(evidence.footerCount, `${route} footer at ${width}px`).toBe(0)
+        expect(evidence.overflow, `${route} overflow at ${width}px`).toBe(false)
+        expect(evidence.headingVisible, `${route} heading at ${width}px`).toBe(true)
+        expect(evidence.ctaOverflow, `${route} CTA clipping at ${width}px`).toBe(false)
+      }
+    }
+  })
+
+  test('utility mobile navigation opens and retains usable links', async ({ page }) => {
+    await page.setViewportSize({ width: 390, height: 844 })
+    await page.emulateMedia({ reducedMotion: 'reduce' })
+    await page.goto(routeUrl('/docs'), { waitUntil: 'networkidle' })
+    await page.getByRole('button', { name: 'Menu' }).click()
+    await expect(page.getByRole('button', { name: 'Close' })).toHaveAttribute('aria-expanded', 'true')
+    await expect(page.getByRole('link', { name: 'Evermind', exact: true })).toBeVisible()
+  })
+})
diff --git a/tests/evidence/public-v4-route-audit.spec.ts b/tests/evidence/public-v4-route-audit.spec.ts
index 9a186af..dc78673 100644
--- a/tests/evidence/public-v4-route-audit.spec.ts
+++ b/tests/evidence/public-v4-route-audit.spec.ts
@@ -18,8 +18,8 @@ const ROUTES: RouteCase[] = [
   { path: '/evermind', variant: 'evermind', motif: 'cinematic', bodySelector: '.cm-canonical-main' },
   { path: '/nevermind', variant: 'nevermind', motif: 'cinematic', bodySelector: '.cm-canonical-main' },
   { path: '/mastermind', variant: 'mastermind', motif: 'cinematic', bodySelector: '.cm-canonical-main' },
-  { path: '/docs', variant: 'docs', motif: 'docs', bodySelector: '.pc-docs-hub' },
-  { path: '/contact', variant: 'contact', motif: 'cinematic', bodySelector: '.cm-contact-page' },
+  { path: '/docs', variant: 'docs', motif: 'docs', bodySelector: '.cm-utility-docs' },
+  { path: '/contact', variant: 'contact', motif: 'cinematic', bodySelector: '.cm-utility-contact' },
   { path: '/privacy', variant: 'legal', motif: 'ledger', bodySelector: ".pc-legal-ledger[data-legal-kind='privacy']" },
   { path: '/terms', variant: 'legal', motif: 'ledger', bodySelector: ".pc-legal-ledger[data-legal-kind='terms']" },
 ]
@@ -36,7 +36,7 @@ const MOTION_SELECTORS: Record<string, string> = {
 }
 
 const CORE_CINEMATIC_ROUTES = new Set(['/', '/evermind', '/nevermind', '/mastermind'])
-const CINEMATIC_ROUTES = new Set([...CORE_CINEMATIC_ROUTES, '/contact'])
+const UTILITY_ROUTES = new Set(['/docs', '/contact', '/privacy', '/terms'])
 
 const REDIRECTS = [
   { from: '/prochat-memory', to: '/evermind' },
@@ -78,22 +78,20 @@ test.describe('site-wide V4 public route evidence', () => {
         expect(response!.status(), `${route.path} returned ${response!.status()}`).toBeLessThan(400)
         expect(normalizedPath(page.url()), `${route.path} redirected unexpectedly`).toBe(route.path)
 
-        if (CINEMATIC_ROUTES.has(route.path)) {
+        if (CORE_CINEMATIC_ROUTES.has(route.path)) {
           await expect(page.locator('.cm-shell')).toHaveCount(1)
           await expect(page.locator('nav.cm-nav')).toHaveCount(1)
-          await expect(page.locator('footer.cm-footer')).toHaveCount(CORE_CINEMATIC_ROUTES.has(route.path) ? 0 : 1)
-          if (CORE_CINEMATIC_ROUTES.has(route.path)) {
-            await expect(page.locator('.cm-shell--core')).toHaveCount(1)
-            await expect(page.locator('.cm-marketing-experience')).toHaveCount(1)
-            await expect(page.locator('.cm-footer iframe')).toHaveCount(0)
-            await expect(page.locator('.pm-navbar,.pc-footer')).toHaveCount(0)
-          }
+          await expect(page.locator('footer')).toHaveCount(0)
+          await expect(page.locator('.cm-shell--core')).toHaveCount(1)
+          await expect(page.locator('.cm-marketing-experience')).toHaveCount(1)
+          await expect(page.locator('.cm-footer iframe')).toHaveCount(0)
+          await expect(page.locator('.pm-navbar,.pc-footer')).toHaveCount(0)
+        } else if (UTILITY_ROUTES.has(route.path)) {
+          await expect(page.locator('.cm-shell.cm-shell--utility')).toHaveCount(1)
+          await expect(page.locator('nav.cm-nav')).toHaveCount(1)
+          await expect(page.locator('footer')).toHaveCount(0)
         } else {
-          const shell = page.locator('.pc-canonical-shell.pc-public-v4')
-          await expect(shell).toHaveCount(1)
-          await expect(shell).toHaveAttribute('data-public-variant', route.variant)
-          await expect(page.locator('nav.pm-navbar')).toHaveCount(1)
-          await expect(page.locator('footer.pc-footer')).toHaveCount(1)
+          throw new Error(`No route-shell contract configured for ${route.path}`)
         }
         await expect(page.locator('main')).toHaveCount(1)
         await expect(page.locator('main')).toBeVisible()
@@ -101,12 +99,6 @@ test.describe('site-wide V4 public route evidence', () => {
         await expect(body, `${route.path} is missing its redesigned body marker`).toHaveCount(1)
         await expect(body).toBeVisible()
 
-        if (!CINEMATIC_ROUTES.has(route.path)) {
-          const scene = page.locator(`.pc-route-scene--${route.variant}`)
-          await expect(scene).toHaveCount(1)
-          await expect(scene.locator(`.pc-route-motif--${route.motif}`)).toHaveCount(1)
-        }
-
         const layout = await page.evaluate(() => ({
           documentWidth: document.documentElement.scrollWidth,
           viewportWidth: window.innerWidth,
@@ -123,7 +115,7 @@ test.describe('site-wide V4 public route evidence', () => {
     await page.setViewportSize({ width: 1440, height: 1000 })
 
     for (const route of ROUTES) {
-      if (CINEMATIC_ROUTES.has(route.path)) {
+      if (CORE_CINEMATIC_ROUTES.has(route.path)) {
         await page.emulateMedia({ reducedMotion: 'no-preference' })
         await page.goto(new URL(route.path, baseUrl).toString(), { waitUntil: 'domcontentloaded' })
         expect(await page.locator('.cm-video__canvas').evaluate(element => getComputedStyle(element).display), `${route.path} canvas should render normally`).not.toBe('none')
@@ -131,6 +123,7 @@ test.describe('site-wide V4 public route evidence', () => {
         expect(await page.locator('.cm-video__canvas').evaluate(element => getComputedStyle(element).display), `${route.path} canvas should stop under reduced motion`).toBe('none')
         continue
       }
+      if (UTILITY_ROUTES.has(route.path)) continue
 
       const selector = MOTION_SELECTORS[route.motif]
       if (!selector) continue

From 83ccb07622860c528df065e326b5532cf9798b07 Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Tue, 29 Sep 2026 11:59:25 +0100
Subject: [PATCH 4/9] fix: preserve contact compatibility and record glass
 evaluation

---
 docs/archive/retired-systems/github-entitlements.md |  8 ++++----
 docs/archive/retired-systems/mailerlite-funnel.md   |  8 ++++----
 docs/automation-routes.md                           |  6 +++---
 docs/implementation-plan.md                         |  4 +++-
 docs/repo-status.md                                 |  2 +-
 docs/roadmap.md                                     |  2 +-
 src/app/(marketing)/contact/ContactPageMarkup.tsx   |  2 +-
 tests/evidence/canonical-chrome-proof.spec.ts       | 10 ++++++++++
 8 files changed, 27 insertions(+), 15 deletions(-)

diff --git a/docs/archive/retired-systems/github-entitlements.md b/docs/archive/retired-systems/github-entitlements.md
index b3f23a2..9bec9cc 100644
--- a/docs/archive/retired-systems/github-entitlements.md
+++ b/docs/archive/retired-systems/github-entitlements.md
@@ -96,7 +96,7 @@ The entitlement state is therefore spread across:
 
 ## Required environment variables
 
-The env contract lives in [environment.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs-public/environment.md).
+The env contract lives in [environment.md](../../../docs-public/environment.md).
 
 GitHub App-specific values:
 
@@ -120,6 +120,6 @@ This doc should not be read as a generic entitlement framework. It documents the
 
 ## Related references
 
-- [integrations.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs/integrations.md)
-- [environment.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs-public/environment.md)
-- [production-lifecycle.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs/production-lifecycle.md)
+- [integrations.md](../../integrations.md)
+- [environment.md](../../../docs-public/environment.md)
+- [production-lifecycle.md](../../production-lifecycle.md)
diff --git a/docs/archive/retired-systems/mailerlite-funnel.md b/docs/archive/retired-systems/mailerlite-funnel.md
index f413b71..efc0adf 100644
--- a/docs/archive/retired-systems/mailerlite-funnel.md
+++ b/docs/archive/retired-systems/mailerlite-funnel.md
@@ -28,7 +28,7 @@ Current behavior:
 
 ## Environment variables
 
-Canonical env documentation lives in [environment.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs-public/environment.md).
+Canonical env documentation lives in [environment.md](../../../docs-public/environment.md).
 
 Current values used by the route (required):
 
@@ -70,6 +70,6 @@ The route does not currently store subscriber records in Prisma.
 
 ## Related references
 
-- [integrations.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs/integrations.md)
-- [environment.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs-public/environment.md)
-- [content-platform.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs/content-platform.md)
+- [integrations.md](../../integrations.md)
+- [environment.md](../../../docs-public/environment.md)
+- [content-platform.md](../../content-platform.md)
diff --git a/docs/automation-routes.md b/docs/automation-routes.md
index 8bf2b88..788e8c6 100644
--- a/docs/automation-routes.md
+++ b/docs/automation-routes.md
@@ -155,6 +155,6 @@ Required env:
 
 ## Related references
 
-- [integrations.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs/integrations.md)
-- [environment.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs-public/environment.md)
-- [development.md](/Users/Office/Repos/Organisation/ProChat/Web/prochat/docs/development.md)
+- [integrations.md](integrations.md)
+- [environment.md](../docs-public/environment.md)
+- [development.md](development.md)
diff --git a/docs/implementation-plan.md b/docs/implementation-plan.md
index f0b1053..ff220cf 100644
--- a/docs/implementation-plan.md
+++ b/docs/implementation-plan.md
@@ -30,7 +30,9 @@ Contact, Docs, Privacy, and Terms use the shared static utility shell: one trans
 - Capture and manually inspect viewport screenshots after motion settles, including top, transition, Section Two, page end, and reduced-motion desktop.
 - Prototype LiquidGlass locally and compare quality, CPU/frame cadence, media scrubbing, reduced motion, and initialization cost. Do not add a dependency or ship the effect without measured benefit.
 
-LiquidGlass evaluation: a real Chrome local prototype using one dynamic scroll-canvas background and one cloned context panel ran at approximately 11–15fps (median frame interval 50–66.6ms) compared with the existing renderer's 16.7ms median cadence. The captured overlay also clipped/reflowed the cloned panel. Decision: do not add the dependency to the scroll path; retain the CSS glass fallback. A separate local Chrome 153 check of the revised cache recorded 89 visible canvas paints in 1.48s (59.99fps), longest paint gap 17.8ms, full 120-frame cache, and three down/up reversals without blanking or console errors.
+LiquidGlass evaluation (2026-09-29, local Chrome 153, 1440×900): the prototype used the real page's scroll-rendered canvas, copied/invalidation only when that canvas repainted, and tested the hero context card and Section Two capability panel independently. Baseline page RAF cadence was 16.62ms average / 16.8ms p95; context-panel prototype was 16.48ms / 16.8ms and capability-panel prototype 16.50ms / 16.8ms. LiquidGlass reported 61fps and 60fps respectively; initialization took 66.0ms and 65.8ms, with 96 changed source frames during each measured scroll sequence and no console errors. This does not reproduce the earlier claimed 11–15fps; that earlier figure is superseded by this instrumented run. Visual inspection rejected adoption: the package added an inset second glass surface/seam to both existing CSS panels and did not improve the composition enough to justify another WebGL context and frame-copy/invalidation path. Keep the existing CSS fallback and add no package dependency. Screenshots: `/tmp/prochat-liquidglass-real-context-measured.png` and `/tmp/prochat-liquidglass-real-capability-measured.png`.
+
+A separate local Chrome 153 renderer check recorded 89 visible canvas paints in 1.48s (59.99fps), longest paint gap 17.8ms, full 120-frame cache, and three down/up reversals without blanking or console errors.
 
 ### C. Validation and release
 
diff --git a/docs/repo-status.md b/docs/repo-status.md
index 9fe66eb..bfba9cc 100644
--- a/docs/repo-status.md
+++ b/docs/repo-status.md
@@ -22,7 +22,7 @@ Generated Docs/Nextra, retired checkout/licensing, Stripe runtime, MailerLite, G
 
 ## Validated release scope
 
-Renderer continuity tests, responsive/reduced-motion and interaction evidence, measured LiquidGlass rejection, production build, 107 repository browser tests, nine focused quality tests, and manual desktop/mobile review are recorded for this release. Chrome 153 locally presented 89/90 tiny-scroll paints at 59.99fps (maximum observed gap 17.8ms), reached all 120 cache frames, and survived three full scroll reversals without blanking or console errors. Local single-run mobile LCP diagnostics: `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, `/contact` 4.21s; all exceed the 2.5s target. Production merge/deployment and remote source housekeeping remain pending. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
+Renderer continuity tests, responsive/reduced-motion and interaction evidence, measured LiquidGlass rejection, production build, 107 repository browser tests, nine focused quality tests, and manual desktop/mobile review are recorded for this release. Chrome 153 locally presented 89/90 tiny-scroll paints at 59.99fps (maximum observed gap 17.8ms), reached all 120 cache frames, and survived three full scroll reversals without blanking or console errors. The real-page LiquidGlass prototype measured 60–61 package FPS and no page RAF cadence regression, but was rejected on visual grounds because it introduced a visible inset/double-glass seam on both candidate panels without enough benefit; evidence is recorded in `docs/implementation-plan.md`. Local single-run mobile LCP diagnostics: `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, `/contact` 4.21s; all exceed the 2.5s target. Production merge/deployment and remote source housekeeping remain pending. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
 
 ## Tracked performance/security notes
 
diff --git a/docs/roadmap.md b/docs/roadmap.md
index f88af56..3a310d3 100644
--- a/docs/roadmap.md
+++ b/docs/roadmap.md
@@ -20,7 +20,7 @@ In scope:
 2. Keep those four routes visually consistent, including Golos display headings, one navigation, translucent capability glass, correct 100svh / 80vh / 100svh geometry, and zero footers.
 3. Bring Contact and Docs into the shared ProChat visual system with restrained/static backgrounds; migrate only shared chrome on Privacy and Terms, preserving legal meaning.
 4. Verify prefers-reduced-motion, 1440/1280/1024/768/430/390/360 layouts, route content, menu and Contact interactions, browser errors, and manually inspected screenshots.
-5. LiquidGlass was tested in a local 1440×900 Chrome prototype on the hero context card and capability panel. It was not adopted: dynamic-canvas performance was about 11–15fps (50–66.6ms median RAF interval) versus 16.7ms baseline cadence, and the overlay clipped/reflowed the composition. Retain the CSS glass fallback; do not add the dependency to the scroll path.
+5. LiquidGlass was tested in local Chrome 153 at 1440×900 on both selected surfaces, with the real scroll-rendered page canvas copied and invalidated on each source repaint. Baseline page RAF cadence was 16.62ms average / 16.8ms p95; the context-card and capability-panel prototypes measured 16.48ms / 16.8ms and 16.50ms / 16.8ms respectively. Package-reported rates were 61fps and 60fps; initialization was 66.0ms and 65.8ms; each run handled 96 changed source frames without console errors. The previous 11–15fps estimate was not reproduced and is superseded. Do not adopt the package: screenshots show an inset, doubled-glass seam on both existing CSS panels without a compelling visual gain. Retain the CSS fallback and avoid adding a WebGL context plus frame-copy/invalidation path. Detailed evidence and screenshots are in `docs/implementation-plan.md`.
 6. Current local Chrome 153 renderer check reached the 120-frame cache and presented 89/90 tiny-scroll paints at 59.99fps, with a longest observed gap of 17.8ms; three full down/up reversals left the canvas painted and produced no console errors. This is a local desktop diagnostic, not a cross-device field guarantee.
 7. Update active implementation docs, validate, review, commit, push, merge through the normal `main` workflow, then verify the live revision and routes. Publication is not complete until that verification passes.
 8. Preserve the protected checkout and unrelated repositories/worktrees. After confirmed production success only, reconcile Mastermind source continuity, indexes, and safely removable branches/worktrees; do not delete unique or unverified work.
diff --git a/src/app/(marketing)/contact/ContactPageMarkup.tsx b/src/app/(marketing)/contact/ContactPageMarkup.tsx
index 3048359..115aa98 100644
--- a/src/app/(marketing)/contact/ContactPageMarkup.tsx
+++ b/src/app/(marketing)/contact/ContactPageMarkup.tsx
@@ -39,7 +39,7 @@ export default function ContactPageMarkup({
         </div>
       </section>
 
-      <section className="cm-utility-contact__form contact-form-panel" aria-labelledby="contact-form-title">
+      <section id="contact-form-card" className="cm-utility-contact__form contact-form-panel" aria-labelledby="contact-form-title">
           <div className="contact-form-panel__header">
               <div>
                 <span>MESSAGE / 01</span>
diff --git a/tests/evidence/canonical-chrome-proof.spec.ts b/tests/evidence/canonical-chrome-proof.spec.ts
index 6692e2d..b09f406 100644
--- a/tests/evidence/canonical-chrome-proof.spec.ts
+++ b/tests/evidence/canonical-chrome-proof.spec.ts
@@ -383,6 +383,16 @@ test.describe('contact page — canonical copy and layout', () => {
     expect(layout.documentWidth).toBeLessThanOrEqual(layout.viewportWidth)
   })
 
+  test('legacy beta contact links still resolve to the contact form', async ({ page }) => {
+    await page.goto(new URL('/contact?topic=memory-qa-beta#contact-form-card', baseUrl).toString(), {
+      waitUntil: 'domcontentloaded',
+    })
+
+    const formSection = page.locator('#contact-form-card')
+    await expect(formSection).toBeVisible()
+    await expect(formSection.locator('form[data-contact-form]')).toBeVisible()
+  })
+
   test('contact desktop screenshot', async ({ page }) => {
     await page.setViewportSize(DESKTOP)
     await page.goto(new URL('/contact', baseUrl).toString(), {

From bc15549f1b975f55fa683f0779aa68cdbf202836 Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Tue, 29 Sep 2026 12:18:22 +0100
Subject: [PATCH 5/9] docs: refresh cinematic release verification count

---
 docs/repo-status.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/repo-status.md b/docs/repo-status.md
index bfba9cc..4906016 100644
--- a/docs/repo-status.md
+++ b/docs/repo-status.md
@@ -22,7 +22,7 @@ Generated Docs/Nextra, retired checkout/licensing, Stripe runtime, MailerLite, G
 
 ## Validated release scope
 
-Renderer continuity tests, responsive/reduced-motion and interaction evidence, measured LiquidGlass rejection, production build, 107 repository browser tests, nine focused quality tests, and manual desktop/mobile review are recorded for this release. Chrome 153 locally presented 89/90 tiny-scroll paints at 59.99fps (maximum observed gap 17.8ms), reached all 120 cache frames, and survived three full scroll reversals without blanking or console errors. The real-page LiquidGlass prototype measured 60–61 package FPS and no page RAF cadence regression, but was rejected on visual grounds because it introduced a visible inset/double-glass seam on both candidate panels without enough benefit; evidence is recorded in `docs/implementation-plan.md`. Local single-run mobile LCP diagnostics: `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, `/contact` 4.21s; all exceed the 2.5s target. Production merge/deployment and remote source housekeeping remain pending. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
+Renderer continuity tests, responsive/reduced-motion and interaction evidence, measured LiquidGlass rejection, production build, 108 repository browser tests, nine focused quality tests (one intentionally skipped), and manual desktop/mobile review are recorded for this release. Chrome 153 locally presented 89/90 tiny-scroll paints at 59.99fps (maximum observed gap 17.8ms), reached all 120 cache frames, and survived three full scroll reversals without blanking or console errors. The real-page LiquidGlass prototype measured 60–61 package FPS and no page RAF cadence regression, but was rejected on visual grounds because it introduced a visible inset/double-glass seam on both candidate panels without enough benefit; evidence is recorded in `docs/implementation-plan.md`. Local single-run mobile LCP diagnostics: `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, `/contact` 4.21s; all exceed the 2.5s target. Production merge/deployment and remote source housekeeping remain pending. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
 
 ## Tracked performance/security notes
 

From 9a930391862827e36036ca56227879c875492d96 Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Tue, 29 Sep 2026 12:31:40 +0100
Subject: [PATCH 6/9] docs: align cinematic site authority and evidence

---
 PRODUCT.md                                    |  4 ++--
 docs/product/HOMEPAGE_MEMORY_FILM_BRIEF.md    |  2 +-
 ...HOMEPAGE_MEMORY_FILM_PRODUCTION_PACKAGE.md |  2 +-
 .../HOMEPAGE_NEXUS_TEMPLATE_ADOPTION.md       | 14 ++++++-----
 docs/product/MEMORY_VISUAL_LANGUAGE.md        | 12 ++++++----
 docs/product/PRODUCT_EXPERIENCE_BRIEF.md      | 16 +++++++------
 docs/product/PUBLIC_CONVERSION_STRATEGY.md    | 12 ++++++----
 docs/product/PUBLIC_PAGE_ARCHITECTURE.md      |  2 +-
 .../evidence/cinematic-final-quality.spec.ts  | 23 +++++++++++++++++--
 9 files changed, 57 insertions(+), 30 deletions(-)

diff --git a/PRODUCT.md b/PRODUCT.md
index 932cdf2..7179ef4 100644
--- a/PRODUCT.md
+++ b/PRODUCT.md
@@ -23,9 +23,9 @@ Mastermind — separate free, open-source planning/orchestration product
 
 ## Public website contract
 
-Canonical routes are `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, `/contact`, `/privacy`, and `/terms`. This cinematic visual contract applies to the first four routes only. The remaining public routes retain their existing page-specific presentation and legal substance; do not redesign them as part of the cinematic release.
+Canonical routes are `/`, `/evermind`, `/nevermind`, `/mastermind`, `/docs`, `/contact`, `/privacy`, and `/terms`. The first four routes use one shared, scroll-driven cinematic template with one full-width navigation, two narrative sections, an intentional 80vh transition spacer, Golos Text display headings, and no footer. The utility routes `/docs`, `/contact`, `/privacy`, and `/terms` use the quieter static cinematic utility shell with one navigation, readable content, and no footer; legal-page meaning remains unchanged.
 
-The home page product sentence is: “Evermind remembers. Nevermind brings context. Mastermind directs the work.” Preserve the distinct product roles and use accurate route-specific calls to action.
+The homepage lead is: “A human-owned system for AI work — from memory to context to controlled execution.” The product-sequence line is: “Evermind remembers. Nevermind brings context. Mastermind directs the work.” Preserve the distinct product roles and use accurate route-specific calls to action. Current implementation and release evidence live in `docs/implementation-plan.md`, `docs/roadmap.md`, and `docs/repo-status.md`; older design/strategy briefs are historical unless those active documents explicitly adopt them.
 
 ## Copy and claim boundaries
 
diff --git a/docs/product/HOMEPAGE_MEMORY_FILM_BRIEF.md b/docs/product/HOMEPAGE_MEMORY_FILM_BRIEF.md
index f27825a..0ba9278 100644
--- a/docs/product/HOMEPAGE_MEMORY_FILM_BRIEF.md
+++ b/docs/product/HOMEPAGE_MEMORY_FILM_BRIEF.md
@@ -2,7 +2,7 @@
 
 **Status:** FROZEN OPTIONAL RESEARCH — not active implementation  
 **Decision date:** 2026-07-18  
-**Superseded by:** `docs/product/HOMEPAGE_NEXUS_TEMPLATE_ADOPTION.md`  
+**Historical design lineage:** superseded by the current website documented in `docs/implementation-plan.md` and `docs/roadmap.md`
 **Phase:** optional future visual experiment only  
 **Purpose:** preserve historical cinematic-memory research; do not use it to direct the current homepage implementation  
 **Strategic authority:** Mind repository  
diff --git a/docs/product/HOMEPAGE_MEMORY_FILM_PRODUCTION_PACKAGE.md b/docs/product/HOMEPAGE_MEMORY_FILM_PRODUCTION_PACKAGE.md
index 471e54f..ffabbf2 100644
--- a/docs/product/HOMEPAGE_MEMORY_FILM_PRODUCTION_PACKAGE.md
+++ b/docs/product/HOMEPAGE_MEMORY_FILM_PRODUCTION_PACKAGE.md
@@ -2,7 +2,7 @@
 
 **Status:** FROZEN OPTIONAL RESEARCH — not active implementation  
 **Decision date:** 2026-07-18  
-**Superseded by:** `docs/product/HOMEPAGE_NEXUS_TEMPLATE_ADOPTION.md`  
+**Historical design lineage:** superseded by the current website documented in `docs/implementation-plan.md` and `docs/roadmap.md`
 **Source brief:** `docs/product/HOMEPAGE_MEMORY_FILM_BRIEF.md`  
 **Target:** historical 12-second scroll-controlled concept only; do not execute without a new explicit roadmap decision  
 **Strategic authority:** Mind repository  
diff --git a/docs/product/HOMEPAGE_NEXUS_TEMPLATE_ADOPTION.md b/docs/product/HOMEPAGE_NEXUS_TEMPLATE_ADOPTION.md
index a92993e..9a5a398 100644
--- a/docs/product/HOMEPAGE_NEXUS_TEMPLATE_ADOPTION.md
+++ b/docs/product/HOMEPAGE_NEXUS_TEMPLATE_ADOPTION.md
@@ -1,11 +1,13 @@
 # ProChat Homepage — Nexus Template Adoption
 
-**Status:** ACTIVE IMPLEMENTATION DIRECTION  
-**Decision date:** 2026-07-18  
-**Owner decision:** adopt the supplied Nexus automation landing-page template as the visual and structural implementation source  
-**Strategic authority:** Mind repository  
-**Design and execution authority:** ProChat repository  
-**Supersedes:** active homepage memory-film and scroll-scrub video direction
+**Status:** SUPERSEDED HISTORICAL DESIGN DECISION — do not use as current implementation direction
+**Decision date:** 2026-07-18
+**Owner decision:** adopt the supplied Nexus automation landing-page template as the visual and structural implementation source
+**Strategic authority:** Mind repository
+**Design and execution authority:** ProChat repository
+**Supersedes:** active homepage memory-film and scroll-scrub video direction at the time; itself superseded by the current cinematic website implementation
+
+> Current authority: `docs/implementation-plan.md`, `docs/roadmap.md`, `docs/homepage-validation-plan.md`, and the current source. This 2026-07-18 Nexus-template decision predates the current Evermind/Nevermind/Mastermind cinematic experience and its Contact/Docs utility shell. Its no-video/no-frame-sequence rules and old product hierarchy are historical evidence only.
 
 ## Decision
 
diff --git a/docs/product/MEMORY_VISUAL_LANGUAGE.md b/docs/product/MEMORY_VISUAL_LANGUAGE.md
index d72796e..77d3073 100644
--- a/docs/product/MEMORY_VISUAL_LANGUAGE.md
+++ b/docs/product/MEMORY_VISUAL_LANGUAGE.md
@@ -1,14 +1,16 @@
 # ProChat Memory Visual Language
 
-**Status:** CANONICAL PUBLIC-EXPERIENCE VISUAL PHILOSOPHY  
-**Decision date:** 2026-07-19  
-**Owner approval:** implementation architecture approved  
-**Strategic authority:** Mind repository  
-**Visual execution authority:** ProChat repository  
+**Status:** historical conceptual visual reference — not the current public website design authority
+**Decision date:** 2026-07-19
+**Owner approval:** implementation architecture approved
+**Strategic authority:** Mind repository
+**Visual execution authority:** ProChat repository
 **Primitive implementation:** `docs/product/MEMORY_ILLUSTRATION_PRIMITIVES.md`
 
 ## Purpose
 
+> This July 2026 Memory-specific visual language remains useful as historical product-concept research. It does not define the current site’s visual system or product hierarchy. Follow the current implementation and roadmap in `docs/implementation-plan.md` and `docs/roadmap.md`, and Mind’s current ProChat brand sources for product truth.
+
 This document defines how ProChat Memory is expressed visually across the public website.
 
 It is not a component API and it is not a page-layout specification.
diff --git a/docs/product/PRODUCT_EXPERIENCE_BRIEF.md b/docs/product/PRODUCT_EXPERIENCE_BRIEF.md
index dba6770..39f6589 100644
--- a/docs/product/PRODUCT_EXPERIENCE_BRIEF.md
+++ b/docs/product/PRODUCT_EXPERIENCE_BRIEF.md
@@ -1,12 +1,14 @@
 # ProChat Public Experience Brief
 
-**Status:** active Phase 1 translation brief  
-**Task:** `PXF-002`  
-**Strategic authority:** Mind repository  
-**Execution authority:** ProChat repository  
-**Next gate:** Nexus-template implementation packet `PXF-003A`  
-**Active visual authority:** `docs/product/HOMEPAGE_NEXUS_TEMPLATE_ADOPTION.md`  
-**Active conversion authority:** `docs/product/PUBLIC_CONVERSION_STRATEGY.md`
+**Status:** superseded historical translation brief — not current product or website authority
+**Task:** `PXF-002`
+**Strategic authority:** Mind repository
+**Execution authority:** ProChat repository
+**Next gate:** Nexus-template implementation packet `PXF-003A`
+**Current website implementation authority:** `docs/implementation-plan.md` and `docs/roadmap.md`
+**Canonical product authority:** Mind, `organizations/prochat/brand/`
+
+> This brief preserves the earlier Memory/Memory-for-QA/Workbench strategy and Nexus design phase. Do not use its product hierarchy, route plan, conversion language, or active-direction labels for the current public website. Current ProChat product truth and site implementation are documented in Mind and the active repository roadmap/implementation plan.
 
 ## Authority
 
diff --git a/docs/product/PUBLIC_CONVERSION_STRATEGY.md b/docs/product/PUBLIC_CONVERSION_STRATEGY.md
index 001f8cd..835cdff 100644
--- a/docs/product/PUBLIC_CONVERSION_STRATEGY.md
+++ b/docs/product/PUBLIC_CONVERSION_STRATEGY.md
@@ -1,14 +1,16 @@
 # ProChat Public Conversion Strategy
 
-**Status:** ACTIVE EXECUTION DIRECTION — MIND RECONCILED 2026-07-19  
-**Decision date:** 2026-07-18  
-**Owner:** Steve Westhoek  
-**Strategic authority:** Mind repository  
-**Execution authority:** ProChat repository  
+**Status:** superseded historical strategy — do not use for current public copy or conversion decisions
+**Decision date:** 2026-07-18
+**Owner:** Steve Westhoek
+**Strategic authority:** Mind repository
+**Execution authority:** ProChat repository
 **Applies to:** homepage, product pages, documentation, contribution paths, contact paths, and future public pages
 
 ## Authority and reconciliation status
 
+> This strategy predates the founder-confirmed 2026-09-11 Evermind/Nevermind/Mastermind product architecture. Its Memory-for-QA/Workbench products, offers, licenses, routes, and CTA rules are historical and are not current public policy. Mind at `organizations/prochat/brand/` is canonical for strategy; the current website execution state is in `docs/roadmap.md` and `docs/implementation-plan.md`.
+
 This document translates the owner's clarified public-growth and service model into website behavior.
 
 Mind reconciliation completed on 2026-07-19.
diff --git a/docs/product/PUBLIC_PAGE_ARCHITECTURE.md b/docs/product/PUBLIC_PAGE_ARCHITECTURE.md
index c0e8419..bda2cb7 100644
--- a/docs/product/PUBLIC_PAGE_ARCHITECTURE.md
+++ b/docs/product/PUBLIC_PAGE_ARCHITECTURE.md
@@ -2,7 +2,7 @@
 
 > Historical plan dated 2026-07-19. Its Memory/Memory-for-QA/Workbench hierarchy, `/memory` routes, page architecture, and acceptance criteria are no longer current. For current route/product guidance see `PRODUCT.md`, `docs/overview.md`, `docs/roadmap.md`, and `docs/implementation-plan.md`. Mind at `mind/organizations/prochat/brand/` remains canonical for product truth.
 
-**Status:** CANONICAL PUBLIC-EXPERIENCE ROUTE AND PAGE-PURPOSE CONTRACT  
+**Status:** superseded historical page-architecture plan
 **Decision date:** 2026-07-19  
 **Strategic authority:** Mind repository  
 **Execution authority:** ProChat repository
diff --git a/tests/evidence/cinematic-final-quality.spec.ts b/tests/evidence/cinematic-final-quality.spec.ts
index e7ce49c..fb72a1b 100644
--- a/tests/evidence/cinematic-final-quality.spec.ts
+++ b/tests/evidence/cinematic-final-quality.spec.ts
@@ -181,9 +181,26 @@ test.describe('ProChat final public-site quality', () => {
       const prototype = CanvasRenderingContext2D.prototype
       const original = prototype.drawImage
       const paintTimes: number[] = []
+      const sample = document.createElement('canvas')
+      sample.width = 24
+      sample.height = 14
+      const sampleContext = sample.getContext('2d')
+      if (!sampleContext) throw new Error('Canvas sampling context is unavailable')
+      const visualFrames: number[] = []
       prototype.drawImage = new Proxy(original, {
         apply(target, thisArg: CanvasRenderingContext2D, args: Parameters<typeof original>) {
-          if (thisArg.canvas === canvas) paintTimes.push(performance.now())
+          if (thisArg.canvas === canvas) {
+            paintTimes.push(performance.now())
+            sampleContext.clearRect(0, 0, sample.width, sample.height)
+            sampleContext.drawImage(canvas, 0, 0, sample.width, sample.height)
+            const pixels = sampleContext.getImageData(0, 0, sample.width, sample.height).data
+            let signature = 2166136261
+            for (let index = 0; index < pixels.length; index += 4) {
+              signature ^= pixels[index] + (pixels[index + 1] << 8) + (pixels[index + 2] << 16) + (pixels[index + 3] << 24)
+              signature = Math.imul(signature, 16777619)
+            }
+            visualFrames.push(signature >>> 0)
+          }
           return Reflect.apply(target, thisArg, args)
         },
       })
@@ -192,7 +209,7 @@ test.describe('ProChat final public-site quality', () => {
         window.scrollTo({ top: 0, behavior: 'instant' })
         await new Promise<void>(resolve => requestAnimationFrame(() => resolve()))
         for (let index = 0; index < 90; index += 1) {
-          window.scrollBy({ top: index % 2 ? -1.5 : 2, behavior: 'instant' })
+          window.scrollBy({ top: 2, behavior: 'instant' })
           await new Promise<void>(resolve => requestAnimationFrame(() => resolve()))
         }
       } finally {
@@ -206,12 +223,14 @@ test.describe('ProChat final public-site quality', () => {
         duration,
         framesPerSecond: duration > 0 ? (paintTimes.length - 1) * 1000 / duration : 0,
         longestGap: Math.max(0, ...intervals),
+        distinctRenderedFrames: new Set(visualFrames).size,
       }
     })
 
     expect(cadence.samples, 'tiny scroll movements repaint the visible canvas').toBeGreaterThan(45)
     expect(cadence.framesPerSecond, `observed paint cadence: ${JSON.stringify(cadence)}`).toBeGreaterThanOrEqual(30)
     expect(cadence.longestGap, `no long gaps during fine scroll: ${JSON.stringify(cadence)}`).toBeLessThan(100)
+    expect(cadence.distinctRenderedFrames, `tiny scroll movements visibly update the image: ${JSON.stringify(cadence)}`).toBeGreaterThanOrEqual(8)
   })
 
   test('four cinematic pages share one clean header and CTA labels stay on a single line at all target widths', async ({ page }) => {

From a6313d1b940a4af2deb7acfda642f603728ba1f7 Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Tue, 29 Sep 2026 12:45:53 +0100
Subject: [PATCH 7/9] test: verify cinematic output through reversals

---
 docs/implementation-plan.md                   |  2 +-
 docs/repo-status.md                           |  2 +-
 docs/roadmap.md                               |  2 +-
 .../evidence/cinematic-final-quality.spec.ts  | 46 +++++++++++++++----
 tests/evidence/playwright.wave1.config.ts     |  2 +
 5 files changed, 41 insertions(+), 13 deletions(-)

diff --git a/docs/implementation-plan.md b/docs/implementation-plan.md
index ff220cf..9af26a3 100644
--- a/docs/implementation-plan.md
+++ b/docs/implementation-plan.md
@@ -32,7 +32,7 @@ Contact, Docs, Privacy, and Terms use the shared static utility shell: one trans
 
 LiquidGlass evaluation (2026-09-29, local Chrome 153, 1440×900): the prototype used the real page's scroll-rendered canvas, copied/invalidation only when that canvas repainted, and tested the hero context card and Section Two capability panel independently. Baseline page RAF cadence was 16.62ms average / 16.8ms p95; context-panel prototype was 16.48ms / 16.8ms and capability-panel prototype 16.50ms / 16.8ms. LiquidGlass reported 61fps and 60fps respectively; initialization took 66.0ms and 65.8ms, with 96 changed source frames during each measured scroll sequence and no console errors. This does not reproduce the earlier claimed 11–15fps; that earlier figure is superseded by this instrumented run. Visual inspection rejected adoption: the package added an inset second glass surface/seam to both existing CSS panels and did not improve the composition enough to justify another WebGL context and frame-copy/invalidation path. Keep the existing CSS fallback and add no package dependency. Screenshots: `/tmp/prochat-liquidglass-real-context-measured.png` and `/tmp/prochat-liquidglass-real-capability-measured.png`.
 
-A separate local Chrome 153 renderer check recorded 89 visible canvas paints in 1.48s (59.99fps), longest paint gap 17.8ms, full 120-frame cache, and three down/up reversals without blanking or console errors.
+Local Chrome 153 recorded 89 visible canvas paints in 1.47s (59.99fps), with 87 distinct sampled outputs and a 17.6ms longest paint gap. A four-route run exercised slow and rapid down/up movement plus three additional reversal cycles per route; each route produced over 500 distinct sampled outputs, remained in the 120-frame cache state, had no transparent visible-canvas samples, and logged no browser errors. The strengthened focused Playwright suite passed 10/10 in system Chrome with H.264 enabled. Bundled CI Chromium skips only the H.264-specific fine-scroll case because that binary has no decoder; bootstrap, continuity, and reduced-motion coverage remain active there.
 
 ### C. Validation and release
 
diff --git a/docs/repo-status.md b/docs/repo-status.md
index 4906016..cb09663 100644
--- a/docs/repo-status.md
+++ b/docs/repo-status.md
@@ -22,7 +22,7 @@ Generated Docs/Nextra, retired checkout/licensing, Stripe runtime, MailerLite, G
 
 ## Validated release scope
 
-Renderer continuity tests, responsive/reduced-motion and interaction evidence, measured LiquidGlass rejection, production build, 108 repository browser tests, nine focused quality tests (one intentionally skipped), and manual desktop/mobile review are recorded for this release. Chrome 153 locally presented 89/90 tiny-scroll paints at 59.99fps (maximum observed gap 17.8ms), reached all 120 cache frames, and survived three full scroll reversals without blanking or console errors. The real-page LiquidGlass prototype measured 60–61 package FPS and no page RAF cadence regression, but was rejected on visual grounds because it introduced a visible inset/double-glass seam on both candidate panels without enough benefit; evidence is recorded in `docs/implementation-plan.md`. Local single-run mobile LCP diagnostics: `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, `/contact` 4.21s; all exceed the 2.5s target. Production merge/deployment and remote source housekeeping remain pending. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
+Renderer continuity tests, responsive/reduced-motion and interaction evidence, measured LiquidGlass rejection, production build, 108 repository browser tests, and manual desktop/mobile review are recorded for this release. The strengthened 10-test focused quality suite passes in system Chrome with H.264 enabled. Chrome 153 locally presented 89 tiny-scroll canvas paints in 1.47s (59.99fps), with 87 distinct sampled outputs and a 17.6ms maximum gap; repeated slow/rapid reverse scrolling across all four routes produced no transparent visible-canvas samples or browser errors. The real-page LiquidGlass prototype measured 60–61 package FPS and no page RAF cadence regression, but was rejected on visual grounds because it introduced a visible inset/double-glass seam on both candidate panels without enough benefit; evidence is recorded in `docs/implementation-plan.md`. Local single-run mobile LCP diagnostics: `/` 2.91s, `/evermind` 2.86s, `/nevermind` 2.86s, `/mastermind` 2.86s, `/docs` 4.05s, `/contact` 4.21s; all exceed the 2.5s target. Production merge/deployment and remote source housekeeping remain pending. The protected dirty checkout `/Users/Office/Repos/prochattools/web/prochat` is preservation-only and must remain byte-for-byte unchanged in its branch, HEAD, status, and tracked diff.
 
 ## Tracked performance/security notes
 
diff --git a/docs/roadmap.md b/docs/roadmap.md
index 3a310d3..1611de1 100644
--- a/docs/roadmap.md
+++ b/docs/roadmap.md
@@ -21,7 +21,7 @@ In scope:
 3. Bring Contact and Docs into the shared ProChat visual system with restrained/static backgrounds; migrate only shared chrome on Privacy and Terms, preserving legal meaning.
 4. Verify prefers-reduced-motion, 1440/1280/1024/768/430/390/360 layouts, route content, menu and Contact interactions, browser errors, and manually inspected screenshots.
 5. LiquidGlass was tested in local Chrome 153 at 1440×900 on both selected surfaces, with the real scroll-rendered page canvas copied and invalidated on each source repaint. Baseline page RAF cadence was 16.62ms average / 16.8ms p95; the context-card and capability-panel prototypes measured 16.48ms / 16.8ms and 16.50ms / 16.8ms respectively. Package-reported rates were 61fps and 60fps; initialization was 66.0ms and 65.8ms; each run handled 96 changed source frames without console errors. The previous 11–15fps estimate was not reproduced and is superseded. Do not adopt the package: screenshots show an inset, doubled-glass seam on both existing CSS panels without a compelling visual gain. Retain the CSS fallback and avoid adding a WebGL context plus frame-copy/invalidation path. Detailed evidence and screenshots are in `docs/implementation-plan.md`.
-6. Current local Chrome 153 renderer check reached the 120-frame cache and presented 89/90 tiny-scroll paints at 59.99fps, with a longest observed gap of 17.8ms; three full down/up reversals left the canvas painted and produced no console errors. This is a local desktop diagnostic, not a cross-device field guarantee.
+6. Current local Chrome 153 diagnostics reached the 120-frame cache and presented 89 tiny-scroll canvas paints in 1.47s (59.99fps); 87 sampled outputs were distinct and the longest gap was 17.6ms. Slow and rapid down/up scrolling plus repeated reversals on all four core routes produced over 500 distinct sampled outputs per route, zero transparent visible-canvas samples, and no console errors. The focused browser suite passed 10/10 in system Chrome with H.264 enabled. This is a local desktop diagnostic, not a cross-device field guarantee.
 7. Update active implementation docs, validate, review, commit, push, merge through the normal `main` workflow, then verify the live revision and routes. Publication is not complete until that verification passes.
 8. Preserve the protected checkout and unrelated repositories/worktrees. After confirmed production success only, reconcile Mastermind source continuity, indexes, and safely removable branches/worktrees; do not delete unique or unverified work.
 
diff --git a/tests/evidence/cinematic-final-quality.spec.ts b/tests/evidence/cinematic-final-quality.spec.ts
index fb72a1b..3e6bf97 100644
--- a/tests/evidence/cinematic-final-quality.spec.ts
+++ b/tests/evidence/cinematic-final-quality.spec.ts
@@ -21,8 +21,14 @@ async function sampleRenderedCanvas(page: Page) {
     const context = scratch.getContext('2d')
     if (!context) throw new Error('Canvas 2D context is unavailable')
 
+    const directionFrames = {
+      slowDown: new Set<number>(),
+      slowUp: new Set<number>(),
+      fastDown: new Set<number>(),
+      fastUp: new Set<number>(),
+    }
     const result = { samples: 0, uncoveredSamples: 0, visibleCanvasSamples: 0, states: new Set<string>() }
-    const read = () => {
+    const read = (direction: keyof typeof directionFrames) => {
       const root = canvas.closest<HTMLElement>('.cm-video')
       const poster = root?.querySelector<HTMLElement>('.cm-video__poster')
       result.states.add(root?.dataset.renderState ?? 'missing')
@@ -43,7 +49,13 @@ async function sampleRenderedCanvas(page: Page) {
       )
       const pixels = context.getImageData(0, 0, scratch.width, scratch.height).data
       let alpha = 0
-      for (let index = 3; index < pixels.length; index += 4) alpha += pixels[index]
+      let signature = 2166136261
+      for (let index = 0; index < pixels.length; index += 4) {
+        alpha += pixels[index + 3]
+        signature ^= pixels[index] + (pixels[index + 1] << 8) + (pixels[index + 2] << 16) + (pixels[index + 3] << 24)
+        signature = Math.imul(signature, 16777619)
+      }
+      directionFrames[direction].add(signature >>> 0)
       result.samples += 1
       // The canvas is intentionally transparent while the poster is the visible
       // fallback. Count a gap only when neither layer covers the transition.
@@ -52,22 +64,33 @@ async function sampleRenderedCanvas(page: Page) {
       }
     }
 
-    const step = async (direction: -1 | 1, count: number, distance: number) => {
+    const step = async (
+      direction: -1 | 1,
+      key: keyof typeof directionFrames,
+      count: number,
+      distance: number,
+    ) => {
       for (let index = 0; index < count; index += 1) {
-        await new Promise<void>(resolve => requestAnimationFrame(() => resolve()))
         window.scrollBy(0, direction * distance)
-        read()
+        await new Promise<void>(resolve => requestAnimationFrame(() => requestAnimationFrame(() => resolve())))
+        read(key)
       }
     }
 
     for (let cycle = 0; cycle < 3; cycle += 1) {
-      await step(1, 42, 5)
-      await step(-1, 42, 5)
-      await step(1, 28, 26)
-      await step(-1, 28, 26)
+      await step(1, 'slowDown', 42, 5)
+      await step(-1, 'slowUp', 42, 5)
+      await step(1, 'fastDown', 28, 26)
+      await step(-1, 'fastUp', 28, 26)
     }
 
-    return { ...result, states: [...result.states] }
+    return {
+      ...result,
+      states: [...result.states],
+      changedFramesByDirection: Object.fromEntries(
+        Object.entries(directionFrames).map(([direction, frames]) => [direction, frames.size]),
+      ),
+    }
   })
 }
 
@@ -111,6 +134,9 @@ test.describe('ProChat final public-site quality', () => {
       expect(sampled.visibleCanvasSamples, `${route} visible canvas samples`).toBeGreaterThan(200)
       expect(sampled.uncoveredSamples, `${route} uncovered background samples`).toBe(0)
       expect(sampled.states.every(state => state === 'bootstrap' || state === 'cache'), `${route} media state stays painted`).toBe(true)
+      for (const [direction, frameCount] of Object.entries(sampled.changedFramesByDirection)) {
+        expect(frameCount, `${route} ${direction} scroll visibly changes the background`).toBeGreaterThan(1)
+      }
       if (canDecodeVideo) expect(sampled.states, `${route} observed cache handoff`).toContain('cache')
     }
   })
diff --git a/tests/evidence/playwright.wave1.config.ts b/tests/evidence/playwright.wave1.config.ts
index 281a5fa..9b0caa2 100644
--- a/tests/evidence/playwright.wave1.config.ts
+++ b/tests/evidence/playwright.wave1.config.ts
@@ -1,6 +1,7 @@
 import { defineConfig } from '@playwright/test'
 
 const isCI = process.env.CI === 'true'
+const browserChannel = process.env.PLAYWRIGHT_CHANNEL
 
 export default defineConfig({
   testDir: '.',
@@ -18,6 +19,7 @@ export default defineConfig({
     : [['html', { outputFolder: './playwright-report', open: 'never' }]],
   use: {
     browserName: 'chromium',
+    ...(browserChannel ? { launchOptions: { channel: browserChannel } } : {}),
     locale: 'en-US',
     timezoneId: 'UTC',
     colorScheme: 'light',

From f83061a80dba7b9b338f2e8618f8880e3333a496 Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Tue, 29 Sep 2026 12:50:26 +0100
Subject: [PATCH 8/9] docs: record reproduced cinematic renderer failure

---
 docs/implementation-plan.md | 2 ++
 docs/roadmap.md             | 2 +-
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/docs/implementation-plan.md b/docs/implementation-plan.md
index 9af26a3..b908868 100644
--- a/docs/implementation-plan.md
+++ b/docs/implementation-plan.md
@@ -16,6 +16,8 @@ Contact, Docs, Privacy, and Terms use the shared static utility shell: one trans
 
 ### A. Renderer and motion
 
+- Root cause reproduced against deployed `c6731140191791e407a2fb6eb27ac96711883e32`: the previous renderer cleared/resized the visible canvas before drawing a candidate and then swallowed draw errors, so a failed candidate could leave that visible canvas transparent. In a production-page browser session, three controlled draw failures took the visible sample alpha sum from 85,680 to 0 while CSS opacity remained 1. This was an injected browser-only failure (not a server mutation); ordinary scroll alone did not reproduce it deterministically. The candidate stages frames offscreen and presents only successful draws; its regression test confirms the last-good pixel checksum is unchanged when a candidate draw fails.
+- The old cache capped sampling at 72 frames with `duration * 10` (about 7.2 frames/second for the 10.04-second asset). The current renderer samples up to 120 frames at 12 frames/second while reducing per-frame width from 720px to 540px, keeping the raw cached-pixel budget approximately constant and improving temporal density. It blends neighboring available frames rather than rounding progress to one frame index.
 - Reproduce slow/fast forward and reverse scroll in a real H.264-capable browser.
 - Prove cache and bootstrap handoffs do not expose blank frames; stage candidate frames offscreen, retain the last good frame after draw failures, blend neighboring frames, and stop painting when settled.
 - Keep temporal sampling at 12fps (120 frames for the current 10.04-second media) within a roughly constant raw bitmap pixel budget by limiting each frame to 540px width.
diff --git a/docs/roadmap.md b/docs/roadmap.md
index 1611de1..838ee50 100644
--- a/docs/roadmap.md
+++ b/docs/roadmap.md
@@ -16,7 +16,7 @@ The release source of truth is `feature/prochat-cinematic-finalization-2026-09`;
 
 In scope:
 
-1. Fix and verify reverse/forward continuity and smooth frame presentation, chapter pacing, and CTA wrapping on the four cinematic product routes.
+1. Fix and verify reverse/forward continuity and smooth frame presentation, chapter pacing, and CTA wrapping on the four cinematic product routes. The deployed `c673114…` failure mechanism was reproduced with controlled browser-only draw errors: clearing the visible canvas before candidate draw allowed a failed draw to leave a transparent surface. The candidate stages offscreen and preserves the last good frame; see `docs/implementation-plan.md` for measurements and test evidence.
 2. Keep those four routes visually consistent, including Golos display headings, one navigation, translucent capability glass, correct 100svh / 80vh / 100svh geometry, and zero footers.
 3. Bring Contact and Docs into the shared ProChat visual system with restrained/static backgrounds; migrate only shared chrome on Privacy and Terms, preserving legal meaning.
 4. Verify prefers-reduced-motion, 1440/1280/1024/768/430/390/360 layouts, route content, menu and Contact interactions, browser errors, and manually inspected screenshots.

From f09f92aab3327b3672df0d593e50040171cc1295 Mon Sep 17 00:00:00 2001
From: Steve Westhoek <info@prochat.tools>
Date: Tue, 29 Sep 2026 14:07:14 +0100
Subject: [PATCH 9/9] fix: self-host ProChat brand fonts

---
 src/app/layout.tsx | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/src/app/layout.tsx b/src/app/layout.tsx
index 90d91d4..c4407d2 100644
--- a/src/app/layout.tsx
+++ b/src/app/layout.tsx
@@ -6,7 +6,6 @@ import UmamiAnalytics from '@/components/UmamiAnalytics'
 import { brand } from '@/lib/brand'
 import { getSEOTags } from '@/libs/seo'
 import { getOrganizationSchema, getWebsiteSchema } from '@/libs/structured-data'
-import { Golos_Text, JetBrains_Mono } from 'next/font/google'
 import localFont from 'next/font/local'
 import { Viewport } from 'next'
 import { ReactNode } from 'react'
@@ -25,16 +24,18 @@ const fontSans = localFont({
   display: 'swap',
 })
 
-const fontBrand = Golos_Text({
-  subsets: ['latin'],
-  weight: ['400', '600', '700'],
+const fontBrand = localFont({
+  src: [
+    { path: '../../public/fonts/GolosText-Regular.ttf', weight: '400' },
+    { path: '../../public/fonts/GolosText-Bold.ttf', weight: '700' },
+  ],
   variable: '--font-brand',
   display: 'swap',
 })
 
-const fontMono = JetBrains_Mono({
-  subsets: ['latin'],
-  weight: ['400', '700'],
+const fontMono = localFont({
+  src: '../../public/fonts/JetBrainsMono-Regular.ttf',
+  weight: '400',
   variable: '--font-mono',
   display: 'swap',
 })