Fixture used, verbatim: https://www.theoffline-club.com —
resulting playbook title: "The Offline Club" (a real global digital-detox brand, HQ
Amsterdam). Not the unrelated "Bali Offline Club" pilot fixture.
Confirmation step — YES, it appeared: "✦ We found more than one face on this site — which one is yours?" with radio options "The part I run" vs "The whole business." I selected "The whole business."
"Bali" scope-leak — NOT a bug. All content is whole-business (Amsterdam HQ, Netherlands/France/Spain retreats, named founders). "Bali" appears once, naming it as one of several example cities in the site's own dropdown — not a narrowing. Reset verified clean (0 rows, 11 tables) before the fresh read.
Gold "Save to Brain" button — corrected: never appeared, and that's CORRECT. I originally flagged this as a bug. It is not — the corrected plan's § Reconciliation confirms the 2-action gold draft band is shipped as dev/vitest-only this PR, not page-reachable in production. CT-PBV2-02 is SUPERSEDED, not FAILed. Retracting the original finding.
| Scenario | Result | Evidence |
|---|---|---|
| CT-PBV2-01 — Empty header | FAIL (blocked step 2) | /brain/book redirects to /brain on a truly empty account |
| CT-PBV2-02 — Draft band (2-action gold) | SUPERSEDED | Shipped as dev/vitest-only this PR — observed behavior matches the reconciled plan exactly |
| CT-PBV2-03 — Confirmed header | PASS | Toast fires, header settles correctly, persists on reload (hydration flash noted, seen 2x) |
| CT-PBV2-04 — Chapter index (10 rows, locks) | PASS | Structure/locks correct; TOC no-op click confirmed non-issue per amended plan |
| CT-PBV2-05 — Empty public chapter | PASS | Correct scaffold + working one-hop Add → |
| CT-PBV2-06 — Empty private chapter | FAIL — launch-relevant | No Coming-soon badge; Add → live and opens a real fillable sheet — inverse of the freeze |
| CT-PBV2-07 — Filled chapter, no scaffold bleed | PASS | Clean rendering on public + private filled chapters |
| CT-PBV2-08 — Read/doesn't-read badges | PASS | Correct split, neutral hairline color confirmed via computed styles |
| CT-PBV2-09 — Recap unnumbered, chapter scale | FAIL (step 1 only) | 24px vs 21px; step 6 (no invented caption) correctly PASSES |
| CT-PBV2-10 — Plain /brain: interim recap card retired | PASS | No old recap card, no TLDR outside book, rail/read-spine intact |
| CT-PBV2-11 — 375px band wrap | PASS | Confirmed-state clean at 375; empty-state reused from CT-PBV2-01's 1440 observation |
| CT-PBV2-12 — Band owns state line, vN not duplicated | PASS | vN chip in band only; cover shows date line without a second chip |
| CT-PBV2-13 — Badge chip v2 contrast | PASS | Code probe: old 40%-alpha rule genuinely gone; live styles match documented v2 pill |
| CT-PBV2-14 — Global negatives + freeze interaction | FAIL | "Not filled in yet" observed live on chapter 03 sub-cards; freeze non-uniform (cross-ref CT-PBV2-06) |
Verdict: MIXED — 9 PASS / 1 SUPERSEDED / 4 FAIL.
failed step: 2 (open /brain/book on a wiped/empty Brain account)
expected: #f-pb-recap-empty — header + state line + dimmed Save-a-copy
actual: client-side replaceState silently redirects /brain/book -> /brain; the
empty-cover book view never mounts (confirmed via console gtag logs)
layer guess: frontend (routing guard) · PRD AC: AC-PLAYBOOK-32, AC-BRAIN-22
failed step: CT-PBV2-06 step 4 (load-bearing negative assertion)
expected: every empty chapter under the shipped co-edit freeze shows NO live Add ->;
private chapters additionally show a Coming-soon badge
actual: private chapters (Revenue confirmed by opening + filling real fields;
reproduced on Direction, Who you serve, How you run across two separate
account states) show NO Coming-soon badge and a fully live, fillable
Add -> — the INVERSE of what the freeze should produce
layer guess: frontend (missing gate) OR the #4321 privateBetaFreeze demote not wired
to these chapters in production · PRD AC: AC-PLAYBOOK-17 (amended)
failed step: 1 (global negative — the string should appear NOWHERE)
expected: "Not filled in yet" never renders anywhere in the book
actual: observed live on 3 sub-offering cards ("Offline Retreats", "Business
Events", "Digital Detox Challenge") under chapter 03 before this
account's offerings were later fully populated
layer guess: frontend — sub-item-level empty state may not be covered by the
fail-line-4 rewrite · PRD AC: AC-PLAYBOOK-17 (amended)
failed step: 1 (Recap title same visual scale as a chapter title)
actual: computed styles — Recap title 24px/weight 500; chapter titles 21px/weight
500 — measurably ~14% larger
layer guess: frontend (CSS token mismatch) · PRD AC: AC-PLAYBOOK-32, AC-BRAIN-141
Data In: https://www.theoffline-club.com, identity gate answer: "The whole business"
Data Out (Recap TLDR, about_story.tldr):
"The Offline Club is a digital-detox movement that brings people together offline through
local city hangouts, multi-day nature retreats (Netherlands, France, Spain), a 24-hour
detox challenge, and curated phone-free events for companies like Heineken and Disney...
Founded in the Netherlands by Ilya Kneppelhout, Jordy and Valentijn Klok; HQ in
Amsterdam."
Broader than the playbook-book-v2 PR's own scope (research-engine content quality, not this rebuild's UI) — flagged separately per Peyton's direct request.
Chapter 02 "Where to find you" (rendered UI): empty scaffold. Raw
brain_profile columns location, locations, maps_url, business_hours,
contact_info, service_area, social_url, sales_channel_urls: all null/empty — a
genuine gap, not a rendering bug.
But discovered_channels DOES have real data (verbatim from REST):
[
{"url": "https://www.instagram.com/theoffline_club/", "platform": "instagram", "confidence": "confirmed"},
{"url": "https://www.facebook.com/p/The-Offline-Club-61558734616406/", "platform": "facebook", "confidence": "looks-like"},
{"url": "https://www.linkedin.com/company/the-offline-club/", "platform": "linkedin", "confidence": "looks-like"},
{"url": "https://luma.com/theofflineclubutrecht", "platform": "luma", "kind": "selling", "confidence": "looks-like"}
]
This surfaces in the Recap's "How you show up" block, never in chapter 02 — a structural observation (different scopes), not a bug.
Ran ONE more real research pass on the same fixture, selecting "The part I run" this time. Second real wallet charge, expected. Wiped Brain first (Sitting C data already captured above); left this comparison read unconfirmed/unsaved.
| "The whole business" | "The part I run" | |
|---|---|---|
| Ch.01 Your business | Digital-detox movement, city hangouts, retreats (NL/FR/ES), corporate events for Heineken/Disney... | "Standard evening hours... walk-ins and groups without prior appointments... unplug, relax, and connect" |
| Ch.03 What you sell | 10 real offerings (retreats, corporate events, licensing, merch, newsletter) | Empty scaffold — nothing extracted |
| Ch.05 Customer Q&A | Empty both times this walk | Filled 3/3: "What time do you close? — 6pm most nights." / "Do you take walk-ins? — Yes, always." / "Can I bring a friend? — Of course." |
| Ch.06 Branding | Rich tone + separate moodboard "THE VIBE" section | One line: "welcoming, informal, and accommodating." No look section. |
| Ch.07 Revenue | Populated once this walk (real income lines) | Empty — also reproduces the CT-PBV2-06 live-Add/no-badge bug a third time |
| Recap "How you show up" | instagram/facebook/linkedin, sells on luma; Guardian/WIRED/LinkedIn/Trustpilot 4.7/251 | Identical, verbatim same line + same press citations |
Finding: the identity-gate choice changes the site-read content dramatically but never narrows the external reputation/social search (identical in both passes — it runs off the business name, not the selected scope). More notably, "The part I run" pass reads like a generic walk-in/retail FAQ (closing time, walk-ins, "bring a friend") that doesn't match anything verifiable on the real theoffline-club.com's public pages, unlike the internally-consistent "whole business" pass. This could be an accurate read of a genuinely thin sub-page (the business has many city chapters), or it could be generic/templated filler when the narrower scope has little to work with — I could not conclusively determine which from this walk alone. Flagging as a research-quality concern for the Brain/research team, not a playbook-book-v2 UI bug.