onbrand-support-full
The WhatsApp assistant answering as the founder: truthful answers from the Brain, human handover when needed, AI-disclosure line on every new chat.
Next: next: CT-OBSF-14 resume then 15..19 + Sitting 3.…
QA runs · visual ledger · generated from test-ledger.md
How the tests ran
Flow Status answers "is this ready to launch" per flow, straight from the queue. History below is every walk that made the ledger, newest first — cards with pictures came from an HTML report that actually inlined the shots; empty frames mean the walk happened and the screenshot stayed in the Chrome session, never saved. Category narrows both sections at once.
Flow status — 103 flows
onbrand-support-peytontianyan
22-turn On-Brand conversation walk on the seeded peytontianyan Brain — needs a real phone + WhatsApp Business connect.
Next: needs a real phone + WhatsApp Business Connect (Meta OAuth) for the 22-turn walk on the seeded peytontianyan Brain — see onbrand-support-full.md 'Distinct sitting' section for full context
activation-wizard
The 3-step Personalize → Connect → Try flow every founder goes through to turn an assistant on.
Next: remaining: CT-ACTIVATION-04 (needs a safely-drainable QA wallet, or Peyton to drain+restore peytontianyan directly) — already ran: CT-ACTIVATION-01 PASS, CT-ACTIVATION-02 PASS(bug:#3396), CT-ACTIVATIO…
pulse-full
Pulse consolidated: statuses, the needs-attention queue, one-click issue fixes, activity counts, and money numbers that agree with Billing everywhere they appear.
Next: next: Section A (CT-PULSE-01..13 Overview/inbox/Health/chip) — flush gates for Section C clear (#4047+#4057 merged).…
wallet-top-up
Spot-check that a real card top-up correctly debits/credits the prepaid wallet on current prod.
Next: remaining: PASS on record (#3513, 2026-08-16) — do not re-walk the click-through.…
booking-flow-full
End-to-end booking: a guest messages WhatsApp, gets the real form, books, the founder marks paid and sees the record on Pulse — plus regression proof of 4 recent booking fixes.
Next: remaining: CT-BOOK-04/06 guest half, CT-ICBF-02b/03/03b, C-2..C-5 (phone-gated); CT-BOOK-08/09 (wallet + Meta-approval-gated); CT-BOOK-07 product ruling (retire vs redirect) — already ran: CT-BOOK-01/…
brain-flow-fixes
Re-verification of the 18-scenario Brain flow fix batch — several scenarios failed and wait on fixes before re-walk.
Next: remaining: CT-BFF-02 needs DOM/computed-style re-inspection; CT-BFF-11 never attempted (slide-over host + activation-wizard Brand Voice z=18500 fold-in); CT-BFF-03/04/08/09 WAIT on a #4340 fix before…
whatsapp-embedded-signup-fresh
A brand-new founder connecting WhatsApp for the first time via Meta Embedded Signup.
Next: remaining: CT-EMBED-02/03 need re-verification by a real (non-automated) browser click, or /debug frontend investigation of handleConnect, before this is trusted as a genuine bug vs.…
brain-offer-field-split
Brain's offer/services field split — does URL-based research resolve to the right business.
Next: superseded 2026-09-02 — folded into brain-research-full (CT-BRRES-91..93 still WAIT on the #4113/#4061 routing call); use /qa run brain-research-full
in-chat-booking-event-retreat
WhatsApp assistant taking an Event/Retreat booking end-to-end inside the chat thread.
Next: SUPERSEDED 2026-09-02 — folded into booking-flow-full (CT-ICBF-02b/03/03b carried as phone-gated); use /qa run booking-flow-full
pulse-activity-display
Pulse activity log rendering — run history, statuses, and the single-Reconnect-verb rule on issues.
Next: superseded 2026-09-02 — folded into pulse-full (CT-PAD leftovers carried); use /qa run pulse-full
pulse-issue-owning-verbs
Whether Pulse issue rows use the correct action verb (Reconnect vs Pending) per the one-verb rule.
Next: superseded 2026-09-02 — folded into pulse-full (plan file was recovered from an unmerged branch — never landed on main); use /qa run pulse-full
pulse-need-attention
The Pulse “needs attention” panel — does it correctly surface the assistant a founder must act on.
Next: superseded 2026-09-02 — folded into pulse-full; use /qa run pulse-full
whatsapp-in-chat-flow
Companion phone-side half of the in-chat booking walk.
Next: SUPERSEDED 2026-09-02 — Round C phone cards folded into booking-flow-full; plan file stamped in place; use /qa run booking-flow-full
beta-coming-soon
Private-beta freeze: generation surfaces show coming-soon veils, live surfaces stay live, and the catalog stays honest.
Next: UNBLOCKED 2026-08-26: #4163 merged 78b90103643e, Cloudflare Pages auto-deploys from main (frontend-only, no backend flush needed).…
brain-generation-full
Paid generation: Brand Voice and FAQ output quality, paid-preview gating, and the budget-gated four-business quality matrix.
Next: next: Sections B/C (FAQ generation + paid-preview) walkable now.…
brain-intake-full
The Share and Teach doors founders drop material into: the composer, attachment chips, and proof that shared material persists into the Brain.
Next: next: Sections A-E walkable now (CT-BRINT-60/61/62 blocked only on account access, not code).…
brain-look-full
How you look: brand colours, mood, and type pulled from a site, Instagram, or dropped images — real, consistent on every surface, never a template default.
Next: next: CT-BRLOOK-05..09 (extract-methods PRIMARY still owed).…
brain-playbook-full
The Brain playbook book: every chapter door opens the right chapter, edits land and stick, and the readiness checklist tells the founder the truth.
Next: remaining: CT-BRPB-01..19 full walk owed (see plan Carried state); NEW CT-BRPB-20..22 (rail overflow cap, PR #4477/#4445) also owed, own dedicated sitting on peyton@mystflo.com — never WP/Ong
brain-start-share-anything
The “Share anything” entry point that kicks off Brain setup from any input.
Next: remaining: CT-SHARESTART-03 / 05 / 09 — already ran: 00/00s/01 PASS; 02–08 MIXED (Sitting B peytontianyan) — next: CT-SHARESTART-03 on #4171 Pages (https://cursor-share-composer-parity.mf-platform-2zd…
lane16-activation-confirm-4199
Verifies the activation confirm-step fix (#4199) on the live app.
Next: PR #4199 merged 2026-08-28 (frontend-only, per its own blocker note), 4+ days ago.…
lane17-brand-voice-verdict-4197
Verifies the brand-voice verdict fix (#4197) — backend flushed, ready to walk.
Next: PR #4197 merged 8551c3a4 + backend-flushed 373cf8bc6964 (brain-ops, activate-assistant, brain-verify-backstop) — walk now
sell-qa-list-redesign
The redesigned sell/QA list surface — layout and behavior after the redesign.
session-fixes-activation-brain
Session fix batch across activation + Brain (#4212/#4199/#4197) — holding for an edge-function flush.
Next: remaining: assistants EF (update-config.ts) confirmed backend-flushed — #4212 fix is an ancestor of the #4969 flush deploy SHA 924b03e0e.…
brain-look-story-3-styles
Three visual-style variants for the Brain “look story” presentation.
Next: remaining: 375px viewport checks; CT-LOOK-06/07 Instagram combination on sitting B; book-edit + index-row deep-checks for sittings B-F; full re-walk on peytonworkprojects@gmail.com (plan's original ac…
brain-research-full
The research engine: site reads that seed the Brain, the identity gate, persistence across sessions, and honest inline errors on dead URLs.
Next: remaining: CT-BRRES-11, CT-BRRES-60-66, CT-BRRES-91-93 (WAIT on Peyton's #4113/#4061 routing call), CT-BRRES-101-104/106/107 (need empty-Brain fixture or incognito tooling), CT-BRRES-110 (needs Google…
brain-attachment-chip
New attachment chip UI when a founder drops a file into Brain.
Next: superseded 2026-09-02 — folded into brain-intake-full; use /qa run brain-intake-full
brain-co-pane-tabs
Split-pane Brain editor — recapturing all 10 chapter tabs after a merged UI change.
Next: drift: plan file never existed on main; chapter-tab coverage lives in brain-playbook-full — use /qa run brain-playbook-full
brain-faq-generation
Auto-generated FAQ content in Brain from a persona-scoped source fix.
Next: superseded 2026-09-02 — folded into brain-generation-full; use /qa run brain-generation-full
brain-identity-faces
Brain's face/identity detection when a founder uploads business photos.
Next: superseded 2026-09-02 — folded into brain-research-full (peytontianyan/Peyton Ong; never Bloom, never WP); use /qa run brain-research-full
brain-ink-air
The new “look hierarchy” build for how Brain organizes and displays a business's visual identity.
Next: superseded 2026-09-02 — folded into brain-look-full (only CT-BRLOOK-22 pill grammar carried; carousel/shelf scenarios retired)
brain-look-unified-card
Consolidated “look” card replacing older separate visual-identity cards.
Next: superseded 2026-09-02 — folded into brain-look-full; use /qa run brain-look-full
brain-our-read-deck
The “read my site” summary deck Brain shows back to the founder.
Next: OBSOLETE 2026-09-02 — Our Read retired as a concept (2026-08-31 playbook-first rewrite); recorded as CT-BRRES-98a in brain-research-full; playbook-first is the current test
brain-paid-preview
Paid preview gating — confirms voice/FAQ previews only unlock after the right payment state.
Next: superseded 2026-09-02 — folded into brain-generation-full (charge-safe ordering preserved); use /qa run brain-generation-full
brain-research
Brain's website/URL research crawl that seeds a business profile automatically.
Next: superseded 2026-09-02 — folded into brain-research-full; use /qa run brain-research-full
brain-research-24v
Variant of the research walk checking the “where you stand” heading and 24-scenario coverage.
Next: superseded 2026-09-02 — folded into brain-research-full (CT-BR24V-25 re-verify carried); use /qa run brain-research-full
brain-share-in-assistant-split
Whether Brain sharing correctly splits from the assistant-specific view.
Next: superseded 2026-09-02 — folded into brain-intake-full; use /qa run brain-intake-full
brain-shelf-lock
Locking behavior on the Brain “shelf” (assistant catalog) view.
Next: RETIRED 2026-09-02 — two-book shelf superseded by playbook-first (AC-BRAIN-21 amended); do not re-walk under any name.…
playbook-book-v2
The redesigned playbook book (v2) — layout, chapters, and edit doors, walked once its PR deploys.
Next: already ran: CT-PBV2-15 render-verified via DB seed — CT-PBV2-15 render PASSED (DB-seed verified); live site-read Luma/social extraction still unproven — needs a non-walled fixture site to close
booking-reset-and-maps-bubble
Nothing to report yet.
brain-doors-progress-line
Nothing to report yet.
brain-teach-staging-tray
Nothing to report yet.
configure-reset-qr-preserve
Nothing to report yet.
faq-six-head-grouping
Nothing to report yet.
header-refresh-auth
Whether the app header correctly refreshes auth state (e.g. after token expiry).
launch-kit-channel-frame-full-flow
End-to-end Launch Kit flow for picking a channel and generating a content frame.
launchkit-caption-pulse-2col
Two-column caption layout for Launch Kit content inside Pulse.
pulse-tab-template
remaining: Fresh view — 8 states never verified end-to-end; #4419+#4421 merged, CF prod deploy confirmed complete
waitl1-premodal-try-gate
Whether the Waitlist+Launch template correctly gates “Try” before activation.
wallet-empty-pause
remaining: Targeted re-walk owed post-fix-wave (2026-09-09), NOT a full CT-WEP-01..07 re-run: (1) escalate a question, then a second answerable question in the same window — AI must answer #2 without…
bali-offline-pilot
Full pilot walk for the Bali Offline Club retreat account.
Next: remaining: Peyton approved wipe-now for Start flow 2026-08-24.…
book-edit
Editing an existing booking record after it's created.
Next: remaining: clickplan not authored yet — no CT- ids until /qa clickplan book-edit lands — already ran: none — status check 2026-08-20: PR #3694 (book-door combination slide, v4.67/v4.68) confirmed MERG…
brain-start-redesign
remaining: CT-BSR-01, CT-BSR-02, CT-BSR-03, CT-BSR-05, CT-BSR-06 need a fresh/zero QA account (not peyton@mystflo.com, not peytonworkprojects@gmail.com, not Ong) with a plan for how to safely reach it…
brain-teach-text-corrections
remaining: CT-06,07,09-16,18-26,28-29 not walked this sitting — already ran: CT-01 PASS, CT-02 FAIL, CT-03 FAIL, CT-04 PASS, CT-05 FAIL, CT-08 FAIL, CT-17 PASS, CT-27 FAIL-as-predicted, CT-30 structur…
budget-cap-alerts
Email alerts when a founder's spend nears their budget cap.
Next: budget-cap email alerts are code-absent under prepaid, not just unconfigured — Resend templates RETIRED 2026-08-18, no NotificationEventType exists for budget_*, send-budget-alert EF no-ops on isPrepa…
co-edit-back-chevron
after PR #4574 merge+deploy (or confirmed live CF preview)
connect-tab-consistency
Visual/behavioral consistency of the Connect tab across app rows.
Next: remaining: CT-CONNECT-05 step5/CT-CONNECT-07 steps3-4 (no Active RTBK01 instance), CT-CONNECT-06 steps2-5 (safety-skipped, live booking assistant), CT-CONNECT-08 steps5-7 (safety-skipped, shared Whats…
demo-ellen-miri-booking-2026-09
vault fixtures needed (KGM PDF/flyers/QRIS image, Offline Club IG screenshots) before Cloud+Chrome can walk it
first-run-chip
The “first run” chip shown to a founder's very first assistant execution.
Next: author /qa clickplan first-run-chip — v4.1 live on Pages
golden-path-persona-brain-matrix
Cross-checks Brain content quality across event-host vs workshop-host personas.
Next: remaining: CT-GOLDEN-04..08 event-host-minimal; full workshop-host-rich sitting; workshop-host-minimal blocked on fixture — already ran: CT-GOLDEN-01 PASS, CT-GOLDEN-02 PASS, CT-GOLDEN-03 PASS
live-email-triggers-deferred
Real transactional emails firing on live triggers (wallet empty, reconnect, auto-recharge).
Next: 4 scenarios now (was 2): CT-EMAIL-03 (wallet_empty_paused) unchanged — waiting on a parallel session's organic $0 drain.…
private-beta-onboarding
Onboarding sequence for a founder invited into private beta.
Next: remaining: CT-PBETA-04(click),CT-PBETA-07,CT-PBETA-09(click),CT-PBETA-12,CT-PBETA-13 — already ran: CT-PBETA-03 PASS,CT-PBETA-10 PASS,CT-PBETA-15 PASS(2026-08-22 live Founder click via banner Link Goo…
reactivation
Reactivating a paused or stopped assistant through its different reactivation modes.
Next: remaining: CT-REACT-01, CT-REACT-04, CT-REACT-07, CT-REACT-09 (both variants), CT-REACT-11-WABA — blocked on peyton@mystflo.com having zero WhatsApp fixture (needs Peyton's manual reconnect).…
stripe-live-topup
after stripe-live-flip-runbook step (e) — STRIPE_TEST_MODE=false flip + CF/EF redeploy (#2430)
booking-personalize-fixes
SUPERSEDED — folded into booking-flow-full; do not run this row, use /qa run booking-flow-full instead
brain-3484
Re-grading a Brain overlay display issue.
Next: superseded 2026-09-02 — folded into brain-playbook-full; use /qa run brain-playbook-full
brain-compounding
superseded 2026-09-02 — folded into brain-research-full (runs after the persistence section on the same tenant, no wipe between); use /qa run brain-research-full
brain-dead-url-error
superseded 2026-09-02 — folded into brain-research-full (fixture bloomretreats.com); use /qa run brain-research-full
brain-ir3-chrome
Brain “IR3” visual-hierarchy check across guest/mobile viewports.
Next: superseded 2026-09-02 — folded into brain-playbook-full (CT-IR3 leftovers carried; guest/375 checks were env-blocked); use /qa run brain-playbook-full
brain-listing
superseded 2026-09-02 — use /qa run brain-playbook-full.…
brain-look-moodboard
superseded 2026-09-02 — folded into brain-look-full (moodboard section still WAIT on brain-look-story-3-styles finishing first)
brain-option-b-share-remount
Alternate (“Option B”) share-and-remount interaction pattern in Brain.
Next: drift: plan file never existed on main; share-remount mechanism covered by brain-intake-full — use /qa run brain-intake-full
brain-periodic-check-retire
drift: plan file never existed on main; coverage lives in brain-playbook-full — use /qa run brain-playbook-full
brain-playbook-branding-index
superseded 2026-09-02 — folded into brain-playbook-full; use /qa run brain-playbook-full
brain-playbook-co-edit
Co-editing behavior on the Brain playbook document.
Next: superseded 2026-09-02 — folded into brain-playbook-full; use /qa run brain-playbook-full
brain-playbook-f9-f33
Organizing pasted content in the Brain playbook (issue #3145).
Next: RETIRED 2026-09-02 — shelf-era plan; history in brain-playbook-full Legacy section.…
brain-research-ingest-parity
superseded 2026-09-02 — folded into brain-research-full (127 parity section keeps its per-scenario READY/WAIT gates); use /qa run brain-research-full
brain-research-persist
superseded 2026-09-02 — folded into brain-research-full (S1 persistence section; frozen-SHA note carried); use /qa run brain-research-full
brain-share-persist
superseded 2026-09-02 — folded into brain-intake-full (CT-SHRP-10/11/12 + residuals carried; account-access blocker noted); use /qa run brain-intake-full
brain-slide-host-dismiss
Dismiss behavior for the Brain slide-over panel when a host closes it.
Next: superseded 2026-09-02 — folded into brain-intake-full; use /qa run brain-intake-full
brain-teach-hours-contact
Peyton ruling 2026-09-15: skip live UI walk on peytonworkprojects/Bali Offline Club account to avoid risking Myriam Parada's concierge-onboarding data this week.…
brain-teach-read
superseded 2026-09-02 — folded into brain-intake-full; use /qa run brain-intake-full
brain-teach-sheet-lock
superseded 2026-09-02 — folded into brain-intake-full; use /qa run brain-intake-full
brain-where-to-find-you
The section of Brain covering a business's “where to find you” details (address, hours, map).
Next: superseded 2026-09-02 — folded into brain-playbook-full (CT-FINDYOU carried); use /qa run brain-playbook-full
event-host-brain
FAQ chip generation specifically for the event-host persona in Brain.
Next: drift: plan file never landed on main (findings file remains); FAQ-chip re-verify carried as a note in brain-generation-full — use /qa run brain-generation-full
faq-per-persona
superseded 2026-09-02 — folded into brain-generation-full (still PARKED there); use /qa run brain-generation-full
guest-empty-wallet
Guest browsing experience with an empty/no wallet.
Next: #3522 second QA account
in-chat-booking-flow
SUPERSEDED 2026-09-02 — folded into booking-flow-full (phone sittings section); do not run this row, use /qa run booking-flow-full
lane18a-booking-token-payment-gate-4204
SUPERSEDED — folded into booking-flow-full; do not run this row, use /qa run booking-flow-full instead
lane18b-booking-savepath-4202
SUPERSEDED — folded into booking-flow-full; do not run this row, use /qa run booking-flow-full instead
lane18c-class-reminder-copy-4200
SUPERSEDED — folded into booking-flow-full; do not run this row, use /qa run booking-flow-full instead
lane18d-qr-comingsoon-4201
SUPERSEDED — folded into booking-flow-full; do not run this row, use /qa run booking-flow-full instead
launchkit-residuals
Leftover empty-state copy and dead nav items in Launch Kit's Active tab.
Next: Active-tab empty copy + dead nav
pulse-billing-metrics-clickthrough
Full Pulse/Billing/Activity-log reconciliation walk.
Next: superseded 2026-09-02 — folded into pulse-full money-sync section.…
pulse-booking-share
superseded 2026-09-02 — content moved into pulse-full Section G (CT-BKSHARE); note CT-BKSHARE-02 goes stale when #4221 merges (CT-QRCARD-01 replaces it); use /qa run pulse-full
pulse-no-feedback-popup
superseded 2026-09-02 — content moved into pulse-full Section F under its native CT-NOFB ids; use /qa run pulse-full
pulse-qr-share-card
superseded 2026-09-02 — content moved into pulse-full Section H (CT-QRCARD), which stays WAIT on #4221 merge; use /qa run pulse-full
pulse-status-attention
superseded 2026-09-02 — stale card-wall plan re-authored inside pulse-full against the current surface; the 2026-08-13 PASS no longer proves anything; use /qa run pulse-full
pulse-status-residuals
Leftover empty-queue / empty-wallet / “Load older” states in Pulse.
Next: superseded 2026-09-02 — residuals folded into pulse-full CT-PULSE-34/38 notes; use /qa run pulse-full
pulse-usage-billing-metrics
Cross-check that Pulse usage numbers match Billing's numbers.
Next: superseded 2026-09-02 — folded into pulse-full (CT-PUBM-11 leftover carried; 01..10 PASS on record); use /qa run pulse-full
pulse-usage-chip-grammar
Copy/grammar correctness on Pulse's usage chips (e.g. “3 runs” vs “3 run”).
Next: superseded 2026-09-02 — folded into pulse-full (recovered from unmerged branch); use /qa run pulse-full
pulse-usage-chips
Visual/behavioral check of the usage chip component itself in Pulse.
Next: superseded 2026-09-02 — folded into pulse-full (recovered from unmerged branch); use /qa run pulse-full
version-activity
Activity-log entries tied to app/version changes.
Next: superseded 2026-09-02 — folded into pulse-full Section I (CT-ALOG tables live in archive/version-activity.md, relocated unchanged); use /qa run pulse-full
Archived — 27 flows (done or confirmed superseded, not deleted)
onbrand-grounding-rewalk
Full 22-scenario grounding re-walk for On-Brand Support — does it answer from real business facts, not hallucinate.
Next: superseded 2026-08-26 — consolidated into onbrand-support-full (Sitting 1 CT-OBSF-00..05)
onbrand-handover-alerts
Confirms the human-handover Slack + Pulse alerts actually fire when a customer needs a real person.
Next: superseded 2026-08-26 — consolidated into onbrand-support-full (Sitting 2 CT-OBSF-10..19)
onbrand-leftover-rewalk
Re-walk the On-Brand WhatsApp assistant's “leftover question” + cron-identity handling after fixes.
Next: CLOSED 2026-08-26 — superseded, folded into onbrand-support-full.md CT-OBSF-04 (provenance note confirms: 'CT-LEFT-03's three-turn card is absorbed here — after the #4060 fix, walk the full 22, not th…
playbook-first
The new playbook-first landing: any paid read lands on the playbook (the old confirm deck is retired), with Save to Brain / re-read in the header.
Next: remaining: CT-PBF-05 (mid-progress crash-resume, not formally timed), CT-PBF-14 (no Instagram screenshot fixture), CT-PBF-15 long-paste truncation (not tested), CT-PBF-18 success+PDF cases (not tested…
bali-offline-sitting-b
Second Bali Offline Club test sitting (duplicate — use bali-offline-pilot).
Next: duplicate flow mint — consolidated back into bali-offline-pilot (same plan doc docs/testing/click-through/bali-offline-pilot.md); use /qa run bali-offline-pilot, not this name
onbrand-config-save-4032-4212
duplicate flow mint — #4032/#4212 already verified today (2026-08-28) under onbrand-support-full Sitting 2: CT-OBSF-10 re-verify PASS after orchestrator redeployed the stale assistants EF (391->393);…
brain-attention-3509
Nothing to report yet.
brain-brand-voice
remaining: Feature is ready to launch — CT-BVOICE-01..15 all done and passing, 2 real bugs found and fixed.…
brain-check-confirm-flip-3558
Nothing to report yet.
brain-look-extract-methods
Different extraction methods Brain can use to pull a business's visual “look”.
Next: remaining: Sitting D CT-LOOKEXT-01..08 (walk only if Peyton asks); CT-LOOKEXT-07/08 stay WAIT (no FFB/Bloom IG vault) — already ran: CT-LOOKEXT-00 PASS, CT-LOOKEXT-10 PASS, CT-LOOKEXT-11 PASS, CT-LOOK…
brain-no-charge-scare
prod BrainPage hash is DV7XaVbL — #3940 live.…
brain-playbook-shelf
Nothing to report yet.
brain-read-vs-manual-seed-rock-island
Nothing to report yet.
channel-frames-post-3704
Nothing to report yet.
connect-funds-doorway
Nothing to report yet.
connection-health-gating
PR #3649 merged 2026-08-18.…
connection-health-gating-3643-3641
Nothing to report yet.
event-booking-try
Try only — not full wizard
event-host
Nothing to report yet.
launchkit-post-activation
rewrite PASS; residuals parked
ledger-proof
Nothing to report yet.
live-email-triggers
remaining: none — superseded — already ran: CT-EMAIL-01 FAIL (pre-#3746 sitting) — Superseded 2026-08-21 by live-email-triggers-postfix-verify — same underlying bug (Resend wallet_topup_receipt).…
live-email-triggers-postfix-verify
remaining: CT-EMAIL-01 re-verify after fix ships — already ran: CT-EMAIL-01 FAIL x3 (PR#3721, PR#3746, today) — root cause now confirmed, not another attempt — root cause FOUND 2026-08-21 via scoped p…
onbrand-date-fallback
remaining: CT-DATEFB-04, CT-DATEFB-05 (the merge proof) — needs Peyton to run the phone card from her personal WhatsApp to Peyton WP; no agent can send WhatsApp — already ran: CT-DATEFB-00 PASS (flush…
onbrand-dense-reply-eval
unblocked 2026-08-24 — flush applied, onbrand-date-fallback ledgered PASS.…
pulse-monitoring
resume only; metrics still partial
whatsapp-in-chat-flow-live-test-plan
remaining: C-2..C-5 need Peyton's phone (Workshop wa.me + Event wa.me + 'what's on?' + bare 'book') — already ran: Workshop TMPL-WSBK01 fresh activate PASS.…
History — 139 walks
54 of 139 have a saved shot
Teach text-correction sweep — 9 of 30 scenarios: ADD reliable, CHANGE/REMOVE broken 3 ways (services/faqs/contact_info)
ADD works cleanly; CHANGE/REMOVE on existing services/faqs/contact_info data fails 3 distinct ways (fabricated duplicates, silent no-op, raw-text paste) confirmed across 5 of 9 walked scenarios — pre-launch blocker for corrections, not additions. Filed #5189 (P0) + #5190 (P1). 21/30 scenarios not walked this sitting, high-confidence predictions recorded per group.
Brain read vs manual seed — Rock Island (Peyton WP, 2026-09-20)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Read extracted 26/26 FAQs (vs manual seed's 6/26) + tiered/residency membership pricing manual skipped + a real live site defect (1-digit-off WhatsApp number); offerings hidden behind a collapsed '10 items ✓ Use' with no rows shown — filed #5135, fix routed to #4715; Instagram + membership-image legs WAIT on Peyton's screenshots
Booking flow full - 2026-09-08: deploy-freshness check for #4569/#4581/#4585, plus CT-BOOK-01/02/04/07/08 status + launch-readiness verdict
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
READY WITH CAVEATS for launch. Confirmed #4569/#4581/#4585 all live in prod (date-picker, photo field, compact spacing, wallet-empty copy). CT-BOOK-01/02 fixtures permanently gone (recommend SUPERSEDED); CT-BOOK-04 reminder-copy half PASS, guest half deferred (no phone); CT-BOOK-07 reproduces open product question (retire vs redirect, no live post-activation edit surface exists); CT-BOOK-08/09 deferred (wallet 0.20 d…
3 earlier runs for this flow
Booking flow full — CT-BOOK-03/05/10-14: real activations + design-parity, P0-class payment_link gate finding
Both fresh activations (Event + Workshop Booking) succeeded; design-parity CT-BOOK-10..14 (R28/R29/R30/D4/R36) mostly PASS. Found a P0-class client-side gap: payment_link required_when gate does not block Personalize-complete state or the Activate CTA when empty (server correctly rejects at final Activate with clean rollback, but founder gets no early warning) -- confirmed on BOTH Event and Workshop Booking. Also con…
Booking flow full — blocked before first scenario (wallet $1.00, live fixtures deleted)
Logged in cleanly via magic-link as Peyton WP but hit two real blockers before any CT-BOOK scenario: wallet balance is $1.00 (plan requires >=$10 for the fresh activations), and the live fixtures ASST-MTATGH0BCL5G/ASST-MTB24SCZ58VZ that CT-BOOK-01/02 depend on have been deleted since 2026-08-27 (likely by the concurrent wrap-20260906-booking-personalize session, live ship-lock). Report: https://mystflo-qa-preview.pag…
Booking flow full — CT-BOOK-03 Event Booking fresh activation (Peyton WP)
Payment-vehicle gate (#4204 fix 3) PASS — clean block + full rollback on empty payment_link. Real bug found + root-caused: unknown-token save validation never fires on copy_flow_intro/copy_payment_ask (still schema-typed textarea, not template_text) — founder can save a broken {{token}} with zero warning. Mitigated for copy_flow_intro by a send-time scrub. One-line schema fix, no code change needed. Fixed the one liv…
Brain start redesign — two-door zero state + rail overflow cap (CT-BSR-01..08)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
Activity export + overflow cap (CT-BSR-08) fully proven end to end — caps at 3, door/sheet count matches, real Google Docs open, wallet unchanged. Guest correctly sees no rail (CT-BSR-04 partial). But 6 of 8 scenarios need a truly empty account — neither QA account we can touch is zero-state, so the brand-new founder's very first screen is still unverified. Found a harmless self-healing render glitch and a real produ…
whatsapp-embedded-signup-fresh — CT-EMBED-01..03
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-EMBED-01 PASS (correct Connect verb, catalog placement). CT-EMBED-02 FAIL: modal opens correctly to idle state, but clicking 'Connect with Meta' opens no popup, fires no Meta-domain network request, and produces no visible state transition across 2 attempts — code review (EmbeddedSignupForm.tsx:181-210) shows none of the 3 coded outcomes (configured-error / sdk-load-error / waiting-popup) match what was observed, …
golden-path-persona-brain-matrix — Sitting 1 (event-host-minimal) CT-GOLDEN-01/02/03
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-GOLDEN-01/02/03 PASS on event-host-minimal (Plus One Society, Peyton WP). Halted before CT-GOLDEN-04 on a concurrent-session tenant collision, confirmed resolved as a coordination miss (mf-platform-cf's own fork), not a product bug.
brain-flow-fixes CT-BFF-12..15 re-walk after PR #4376/#4381 (DOCX central-directory fix)
Re-walk of CT-BFF-12..15 after PR #4376 deployed (1fd5ce9c2, brain-ops v132). CT-BFF-12 PASS: multi-font Google-Docs-exported PDF extracted verbatim (3/3 Q&A pairs matched pdftotext source). CT-BFF-13 FAIL(backend)->PASS(surfacing) then PASS after fix: DOCX upload showed honest inline error instead of generic banner, but root cause was a real bug (findZipEntry trusted DOCX local ZIP header instead of central director…
1 earlier run for this flow
Brain flow-fixes batch walked live: 3/5 PRs clean (OAuth reconnect #4346, Keep-checked retirement #4146, TLDR render #4343), 2/5 broken in prod (chapter-fork #4344, file uploads #4341) + CT-BFF-16 contradicts Peyton's own report
3 of 5 fixes verified clean in prod (Google OAuth Add-to-Brain reconnect, the new source-picker menu, Keep-my-brain-checked fully retired). 2 of 5 do not work: the Teach/Share door still silently misroutes a multi-chapter business (filed #4348, P0), and EVERY file upload (PDF/DOCX/TXT) fails outright (reopened #4327, P0). Also reopened #4340 (Google Doc content never actually read, P1/P2) and filed #4349 (QA reset sc…
brain-research-full continuation — CT-BRRES-06/10/90/100-107
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
Continuation sitting: CT-BRRES-06 MIXED (safe, generic copy — evidence on #4424), CT-BRRES-07 SKIP (no fixture), CT-BRRES-10 WARN (real gap, evidence on #4424), CT-BRRES-90 PASS, CT-BRRES-94 SKIP (no fixture), CT-BRRES-100 DIVERGENCE (real finding — AC-BRAIN-91 vs shipped isGuidedShareComposer split), CT-BRRES-101/102/103/104/106/107 BLOCKED (no empty-Brain fixture / no incognito tooling), CT-BRRES-105 attempted-not-…
1 earlier run for this flow
brain-research-full — partial walk 2026-09-03 (Sections A/B/C-leftovers/D/F/H/E2)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
~60% of the plan walked. Core research/identity/persistence/compounding loop is close to solid. Two real bugs filed: P0 #4417 (cross-business content contamination in the private reflection notes right after a fresh paid read, self-corrected on next teach) and #4414 (the day-one 'read whole site or tell me yourself' prompt never appears live). A real overcharge on established-Brain second-URL shares tied to the alrea…
CT-PBF-16/17 Google Doc/Sheet extraction verification
Add to Brain (Google Doc/Sheet) runs cleanly but silently drops structured content - FAQ pairs and workshop prices never become organized drafts, only a vague summary paragraph. Filed as backend bug.
3 earlier runs for this flow
playbook-first expanded intake-channel walk (CT-PBF-00..15,18..22): core site-read cycle works, 3 significant regressions found (Teach-share content invisible on established Brain, both /brain/read residual doors dead, nameless day-one share never asks identity)
Core website-read-to-playbook cycle works end to end with real content and real billing. 3 significant regressions found: paid Teach-share research becomes completely invisible on an established Brain (no confirm gate ever appears); both doors to the old research-lookback page are dead; a nameless day-one share never asks who the business is. Also confirmed a QA-tooling gap (reset script misses brain_read_drafts/brai…
playbook-first
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-PBF-16 FAIL (Add to Brain never reads the Doc, silently runs a paid web-research pass instead, confirmed via DB), CT-PBF-17 BLOCKED (Sheet never offered in picker); 21 scenarios remain
Playbook-First Landing — deck retirement (a-seam / #4318 / c-header)
0/23 walked — blocked before first click by a QA-tooling bug (mintMagicLink in scripts/prove/lib.ts self-consumes the magic-link OTP before Chrome can use it, confirmed via otp_expired console error). Account is wiped and ready; needs a login-link code fix to resume.
playbook-book-v2 round 2 (CT-PBV2-15 + re-checks)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-PBV2-15 BLOCKED on fixture (theoffline-club.com bot-walled, no manual social-entry field, avoided wiping shared fixture) — honesty-boundary half PASS, render half unproven live (vitest-only proof stands). CT-PBV2-09 (#4385) re-confirmed FAIL via computed style (24px vs 21px). CT-PBV2-14 step 1 (#4383) inconclusive — chapter 03 has zero offerings on this account, not re-verifiable. CT-PBV2-01 (#4384) not re-tested,…
3 earlier runs for this flow
Designed Playbook v2 (corrected, CT-PBV2-01..14 + confirmation-scope comparison)
9/14 checks pass; header works as designed (earlier false alarm retracted). Real issues: private chapters allow live editing with no Coming-soon lock (launch-relevant), empty-book cover redirects away, a retired 'not filled in yet' label still appears, Recap heading slightly oversized. Extra: a second paid comparison pass shows the site-read scope choice ('whole business' vs 'the part I run') produces very different …
Designed Playbook v2 header/index/scaffold/badges/recap (corrected, CT-PBV2-01..14)
Corrected re-score: the 3-state header IS working as designed (the original 'header broken' finding was a false alarm from a stale test-plan copy, now fixed). Real issues: private chapters allow live editing with no Coming-soon lock (launch-relevant), the empty-book cover redirects away instead of rendering, a retired 'not filled in yet' label still appears on partial offerings, and the Recap heading is slightly over…
Designed Playbook v2 header/index/scaffold/badges/recap
Structure (chapter index, scaffold placeholders, read-labels, save transition) works, but the 3-state header never shows the gold Save-to-Brain/re-read row for a pending draft, the true empty-book state is unreachable, TOC rows don't jump to chapters, and private chapters wrongly allow editing with no Coming-soon badge -- not ready to call done.
On-Brand Support — live phone incident response + fix re-verification (2026-08-28/29)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Live phone sitting (not Chrome) 2026-08-28/29: production incident found+fixed+re-verified — tone/cadence fix #4264, duplicate-reply dedup #4265, P0 outage fix #4272, all confirmed working live post-deploy. Validates CT-OBSF-04/20/21 behavior. Plan's own remaining scope (CT-OBSF-14..19, Sittings 3-4) unchanged and still open.
5 earlier runs for this flow
On-Brand Support consolidated re-walk -- Sitting 2 CT-OBSF-10..13 re-verify (post-redeploy)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Root cause was a stale EF deploy, not the code fix -- after the orchestrator redeployed assistants, the Personalize save + Slack toggle + Connect picker all check out clean (verified with page reloads, not just optimistic UI). CT-OBSF-14 interrupted by a browser disconnect, resuming.
On-Brand Support consolidated re-walk — Sitting 2 CT-OBSF-10 (#4032 recurred post-merge)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
The exact Personalize-save bug the merged fix (#4212) was supposed to close recurred on the very first toggle save attempted after merge -- same FunctionsHttpError signature. Sitting 2 stopped immediately per the plan's own instruction.
On-Brand Support consolidated re-walk — Sitting 1 grounding (CT-OBSF-00..05)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
5 of 8 previously-broken grounding turns now answer correctly; 2 still fail (one is a new regression) and 1 was never asked live -- real progress, not launch-ready. Sitting 2 stays blocked on an unmerged fix.
onbrand-support-peytontianyan
Peyton-directed run on peytontianyan@gmail.com (distinct from locked onbrand-support-full.md account, peytonworkprojects@gmail.com). Brain wiped + reseeded (Salt & Cedar, 8/10 chapters, via real Teach-your-Brain paste flow). Activation wizard walked through Personalize (Warm Host + AI identity line confirmed live and correctly reading Salt & Cedar in the preview) and Connect (WhatsApp Business shows Connect, not conn…
onbrand-support-full — gate check BLOCKED (Brain in live use + #4032 gate missing)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-OBSF-00 gate check only: wallet/connects/account verified, TMPL-9P3X not yet active, Brain still Clay Room. Sitting-1 fix receipts (#4060/#4069/Warm Host #4165) confirmed applied. #4032 Personalize-save fix has no merged PR (issue open, #4212 unmerged) — blocks Sitting 2. Brain reseed held: Workshop Booking shows 'Chatting now' on the same tenant Brain right now — reseeding would corrupt a live concurrent conversa…
brain-research-persist — HQ/Bali disambiguation continuation, BLOCKED on #4223/#4224
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Attempted to re-verify #4183-class disambiguation (HQ vs Bali chapter) via the research/Teach flow. Blocked by two newly-found bugs: #4223 (P0, address-redaction filter strips place names like 'Bali' from pasted text before disambiguation logic can see them) and #4224 (P1, URL reader fails on a live, curl-reachable site with a standard www redirect). No bad data written — corrupted draft declined before confirm, no p…
connect-tab-consistency: first full execution
Two-standalone-controls row shape holds across all 4 assistants pre/post; found 2 real UI gaps (AI Model leaks a disabled Switch-account item; WAITL1 multi-app Disconnect is missing its disambiguation ellipsis) plus a likely #3120 fix already live; width sweep + RTBK01 post-label + the risky multi-app disconnect click still need a dedicated re-run
Workshop Booking real WhatsApp phone round (TMPL-WSBK01) — DB reconciliation
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Real WhatsApp booking on Workshop Booking billed correctly once ($1.75 wallet debit, charged, no double-charge) and the schema DOES have paid_at/confirmed_at gating columns — but assistants.cost_this_week stays $0.00 against a real matching wallet charge (display drift, AC-BILL-02), and whether the guest's 'confirmed' message arrived before or after the host's Mark Paid tap is unresolved from DB alone (needs Peyton's…
3 earlier runs for this flow
whatsapp-in-chat-flow: Workshop activation + Round C-1
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Activated a brand-new Workshop Booking assistant alongside the already-Active Event Booking assistant — both now run concurrently on Peyton WP; confirmed on Pulse at desktop and mobile. No repeat of this morning's brand-voice activation bug. Real-phone half (Round C-2..C-5) still needed.
In-chat booking Workshop Activate retest (Peyton WP) — PR #3706 fix confirmation
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-ICBF-01 Workshop Activate now succeeds live on Peyton WP after PR #3706 (waba-credentials pgcrypto cipher fix) merged+deployed. Confirmed in Pulse: Workshop Booking (Wheel-Throwing for Beginners) Active. Event/Retreat personas + CT-ICBF-01b not yet run — out of scope for this confirmation pass.
In-chat booking three personas (Peyton WP)
Workshop Personalize+Connect fully verified vs plan; Activate blocked by live WhatsApp/Meta credential rejection contradicting 'Healthy' status on /connects. Fixed reset-test-account.sh generation_runs FK gap live. Found wizard-state-lost-on-reload gap. Event/Retreat not started (same blocker).
whatsapp-in-chat-flow: Workshop activation + Round C-1 (queue-key correction)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Corrects the queue key this progress landed under (whatsapp-in-chat-flow, not in-chat-booking-flow) — same walk as the prior row: fresh Workshop activate + Round C-1 PASS.
Share anything persists to all three surfaces (ACCEPTANCE TEST for #4092)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Acceptance test for #4092 P0 fix: PASS (CT-SHRP-02 flagship POST-FIX). Plan's stated job is done. Residual leftover (CT-SHRP-06 no-fixture, CT-SHRP-07 PDF, CT-SHRP-04 GAP) and the 3 bugs the walk surfaced (#4183, #4184, #4185) are now tracked as their own issues + private-beta lanes 12A/12B/12C on #4159 — closing this plan rather than holding it open for a re-sitting that duplicates lane tracking.
1 earlier run for this flow
brain-share-persist — #4092 flagship (CT-SHRP-02) POST-FIX PASS
POST-FIX walk (PR #4160 merged): the #4092 acceptance test passes — business name + description now persist to brain_profile, survive refresh + full re-login, charges match ($0.854691). Services stays empty as expected (separate known gap). Our Read note confirmed NOT minted on share-origin (documented gap, not new). Report: https://mystflo-qa-preview.pages.dev/qa/brain-share-persist-report-2026-08-26.html. Leftover:…
Look story 6-builder harvest — PR #4118 re-walk (redirected to peytontianyan, wiped) — 6/6 sittings FAIL
Redirected per Peyton's live call to peytontianyan@gmail.com (wiped) instead of the plan's peytonworkprojects (dirty, no-wipe). Walked CT-LOOK-00/05 (PASS) + CT-LOOK-02 sittings A-F (harvest) + CT-LOOK-03 sitting A (book/index interior). Result: 6/6 sittings show real defects — confirmed sheet-vs-book Signature SYNC bug on A; Trustpilot third-party-widget contamination on B; wrong-CSS-role promotion with correct valu…
Brain Start Share anything — 3-chief launch-readiness panel: DO-NOT-PASS, corrected root cause
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Post-walk launch-readiness evaluation: Chief Product, Chief Designer, Chief CTO each independently reviewed the CT-SHARESTART-08 finding against real code (Product+CTO on Opus after a Fable limit fallback). ALL THREE returned DO-NOT-PASS. Chief CTO corrected the root-cause diagnosis: the paid draft IS saved durably (ADR-023 by design, no data loss, no refund concern) -- the actual bug is the playbook shelf gate (Play…
7 earlier runs for this flow
Brain Start Share anything — CT-SHARESTART-08 reclassified P0 (Chief Product diagnosis, not a UI gap)
Correction to the 2026-08-24 sitting: what read as 'Review drafts on your playbook does not navigate / no playbook reachable' is NOT a missing UI screen. Chief Product (fable, opus fallback after a Fable limit) confirmed the playbook UI is fully shipped (PlaybookShelf.tsx, PlaybookBookPage.tsx, /brain/book, tests) and gates on a saved business_name/description/services in brain_profile — this Share-anything path neve…
Brain Start Share anything — CT-SHARESTART-07/08 complete, wallet funded, plan DONE
Peyton funded the peytontianyan wallet herself ($25, new Visa card via /billing — I never touch card fields). Walked the final two scenarios: CT-SHARESTART-07 (ONE Save) staged all 4 real vault fixtures (profile + 3 posts), Save fired a $0.20 organize/vision charge, sheet closed, and the identity gate correctly extracted 'The Offline Club' as the business name straight from the shared IG content. CT-SHARESTART-08 (id…
Brain Start Share anything — CT-SHARESTART-02..06 on peytontianyan (empty-Brain, no wipe needed)
Peyton redirected Sitting B onto the already-signed-in peytontianyan account instead of peytonworkprojects@gmail.com — it turned out to already be a fresh empty-Brain day-one account, so NO wipe was needed and the bali-offline-pilot collision was sidestepped entirely. Walked CT-SHARESTART-02 through 06, all PASS: signed-in empty-Brain door+sheet, exact left-pane copy (title/sub-line/5 rows/subtitle), layout (guidance…
Brain Start Share anything — CT-SHARESTART-00s empty shelf, arrow slide, Our read stands (local Chrome)
CT-SHARESTART-00s PASS: See what you'd get shelf renders (4 example covers, kickers Example, persona titles, business credits), → arrow slides playbooks only (dark What we understood card stays pinned outside the row), tapping the standing card opens the Our-read example deck with no share sheet / no charge. Re-confirmed CT-SHARESTART-00 door copy exact match while at it. Local Claude-in-Chrome resize_window floor is…
Brain start Share anything — wipe-now approved, 02-08 BLOCKED infra
Peyton approved wipe-now. Start flow already shipped (#3992; 74 unit tests; prod BrainPage-BRR5N5HB + new door). Vault inbox has Offline Club profile + posts. This Cloud VM cannot wipe or sign in (no SUPABASE_DB_URL / QA_USER_PASSWORD). 00/01 still PASS. Leftover 02-08. Gallery: https://mystflo-qa-preview.pages.dev/qa/brain-start-share-anything-report-2026-08-24.html
Brain start Share anything — door live, guest gate PASS
New door is live (BrainPage-BRR5N5HB). Guest tap opens the account gate. Empty-Brain 02-08 wait on bali-offline-pilot DONE before wipe. https://mystflo-qa-preview.pages.dev/qa/brain-start-share-anything-report-2026-08-24.html
Brain Start Share anything — CT-SHARESTART-00 deploy gate
CT-SHARESTART-00 BLOCKED: prod still BrainPage-DlCD1-0Q.js old door ('Screenshot, PDF, or a Google Doc — Share anything →'). #3992 not on CF prod. Report: https://mystflo-qa-preview.pages.dev/qa/brain-start-share-anything-report-2026-08-24.html
Brain IR3 chrome (PR #4116+#4132) — branding-only mounts IR3
CT-IR3-01/02 PASS on prod (peytontianyan, 13% branding-only): two books, hairline examples, Teach on right rail, Where you stand (7 of 8 required), Activity on rail; screenshot bug does not reproduce on fresh navigation. CT-IR3-00 (guest) and CT-IR3-03 (375) BLOCKED by this Chrome session's infra (no incognito capability; resize_window is a no-op, confirmed via window.innerWidth) — not a product finding. URL: platfor…
How you look — 3 start doors (research/IG/both) — PRIMARY (CT-LOOKEXT-00,10,11,12,13)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
All 5 PRIMARY scenarios PASS. Gate confirmed #4075+#4085 live in prod. Two wipes verified clean+wallet-safe. Research-only, research+IG, and IG-only doors persist correctly with scope wall verified in DB. Minor bug: vibe chip mislabels source on research-only door. FOLLOW-UP (same day): live 3-site differential test found the CSS-harvest color pipeline (_shared/style-harvest.ts) returns colors unrelated to the real s…
Brain offer-field-split — Luma URL research resolves wrong site (CT-OFFERSPLIT-01)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-OFFERSPLIT-00 PASS. CT-OFFERSPLIT-01 FAIL — Luma URL paste researches theoffline-club.com (parent brand site), not the Luma calendar; zero services[] rows produced; $2.50 real charge. Filed #4097 for build-side brain-scrape trace. CT-02/03 not exercisable, CT-04 skipped (no fixture). Report: docs/testing/testing-artifacts/brain-offer-field-split-report-2026-08-25.html
onbrand-grounding-rewalk checkpoint — CT-GROUND-04 turn-12-equiv PASS, turns 5/10-equiv reclassified #4022->#4060 (2026-08-25b)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Partial re-walk via 4 real WhatsApp messages from Peyton's phone (Chrome cannot send WhatsApp). Turn 12-equiv (room price diff) now PASSES — currency-aware grounding.ts fix (PR #4071) confirmed working. Turns 5/10-equiv (schedule, hold-a-spot) still FAIL the gating bar — no longer flat-deny (that part of #4022 fixed) but now hedge/over-answer/escalate instead of stating known facts — reclassified to #4060's bug class…
1 earlier run for this flow
On-Brand grounding re-walk — CT-GROUND-04 (22-turn phone card, post #4037/#4039/#4040 flush)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
19/22 gating turns pass (86.4%), 5/8 previously-FAIL turns now fixed (location, deposit, balance-due, cancellation, memory-recall); turns 5/10 still conflate empty live calendar with no schedule/no hold policy, turn 12 drops a fact already given earlier in-session — new consistency issue #4060 filed, #4022 commented not closed; human-likeness gap flagged; Slack founder-reply leak (#4058) referenced, not re-filed; ALS…
Backfill: CT-BR24V-10 (Bloom, BLOCKED dead fixture) + CT-BR24V-20 (Offline Club, PASS, real billing evidence)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-10 BLOCKED: bloomretreats.com is DNS-dead (confirmed via dig against 3 resolvers), not a product bug -- no charge. CT-20 PASS: Offline Club paid research completed, one picker (AC-BRAIN-95), correct face-scoped output, billing directly verified via /billing/history (Brain deliverable -$2.50 + AI usage, wallet $25.00->$21.54). This is the missing baseline an independent CTO review flagged as an unverifiable premise…
1 earlier run for this flow
Signed-in refuse, Share-anything completion, rail contract, identity correction (CT-BR24V-21..25)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-21/22 PASS (signed-in Notion + Google Doc refuse, correct copy, no charge). CT-23 PASS mechanically, real product gap found (Instagram paste bypasses voice/look vision-read on the unscoped share door) -- fix approved by Peyton, routed to a separate build handoff. CT-24 discovery: 'Not quite' correction path confirmed real and well-built, no charge, fixture kripalu.org. CT-25 partial: Activity PASS with real timest…
On-Brand handover alerts — Personalize config save (CT-HAND-01)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-HAND-00 PASS (flush confirmed applied despite backend-flush-queue.sh list bug). CT-HAND-01 BLOCKED: Personalize Save Changes fails 4/4 attempts with 'Update failed' toast + console FunctionsHttpError non-2xx on update-config EF. Filed #4032 (prime suspect: new Slack-stamp-into-workflow_instances block in update-config.ts). CT-HAND-02..09 not run — no WhatsApp/Slack turns spent. Session tab group also dropped mid-r…
Brand Voice — consolidated 7-sitting status (2026-08-16 → 2026-08-22)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Ready to launch. 2 real bugs found+fixed (stale voice summary after regenerate; typed reply silently lost on sheet close, 2 code paths). Extraction confirmed grounded/specific on real signal (Offline Club regenerate), correctly gated at 0% signal. 1 cosmetic copy bug filed unfixed. Open: CT-BVOICE-16 (2nd-business spot-check, needs Peyton go-ahead for real $5 charge) — nice-to-have, not a blocker.
13 earlier runs for this flow
Brand Voice — consolidated status (no new walk; #3978)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Plain-English consolidation of 2026-08-16..22 sittings already ledgered. No new Chrome walk. Ready to launch; CT-BVOICE-16 still a nice-to-have. Row backfills #3978 so qa-assert-ledger matches the findings file on main.
Brain Voice/Tone — CT-BVOICE-13/14/15 post-fix re-verify (PRs #3840, #3842, #3845)
All three post-fix scenarios PASS. CT-BVOICE-13 (#3840): real $5 regenerate confirmed the fix — new specific one-liner replaces stale 'friendly'. CT-BVOICE-14 (#3842): all 3 composers restore/clear cleanly, no lost text. CT-BVOICE-15 (#3845): stale-voice nudge live on both surfaces, non-blocking, no charge on open. Closes the CT-BVOICE-07..15 signal-completeness ladder for this wave.
brain-brand-voice CT-BVOICE-12 final re-verify — PASS, #3813 closed for real
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-BVOICE-12 re-run post PR #3828 (the real #3813 fix, after #3821's first attempt missed the REFINE-view composer). Confirmed: localStorage draft write fires on the REFINE-view 'Add a real reply you wrote' composer, survives soft reopen AND hard reload, clears cleanly on successful save. Full CT-BVOICE-07..12 signal-completeness ladder now closed: 2 real bugs found and fixed (#3781, #3813), #3813 required a second i…
brain-brand-voice CT-BVOICE-11/12 post-fix re-verify
CT-BVOICE-11 PASS (Branding row now correctly refreshes after Save, survives hard reload - #3781 fix confirmed solid). CT-BVOICE-12 FAIL (typed sample still lost on close, exact original repro - #3813 reopened with new evidence, fix's persistence layer not reaching the REFINE-view composer). Report: docs/testing/testing-artifacts/brain-brand-voice-ct11-12-verify-2026-08-21.html
Brain Voice/Tone — CT-BVOICE-08/09 extraction + unconfirmed-edit ladder steps
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-BVOICE-08 PASS: real mystflo.com scrape produced a specific voice DRAFT ('Direct, punchy, founder-empathetic...Pay per run. Not per month.') but gather sheet + generate lock stay unchanged from 0% — lock gates on CONFIRMED rows, not drafts. CT-BVOICE-09 FAIL: typed founder sample silently lost on sheet close, not draft-saved — filed issue #3813. Only cost was the $2.50 research scrape; no generate fired.
Brain Voice/Tone — CT-BVOICE-07 0% baseline ladder step
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
0% signal account: tone pre-fill is generic 'Warm & casual' template, matches Peyton's complaint pattern exactly — expected floor, not a bug. New finding: generate action is LOCKED (not just priced) until name+what-you-do+one-offering is added — stronger guardrail than plan assumed. No AI charge fired.
Brain Voice/Tone — CT-BVOICE-10 authorized Regenerate + Save (Peyton WP, rich report)
Fired 2nd real $5.10 Regenerate on the primary (signal-rich) QA account and Saved it for real per Peyton's go-ahead ('because it is us, we fix it now'). New output specific/grounded again (refund policy, ticket-vs-RSVP, Bali audience). Confirmed live after hard reload. NEW BUG found + filed: Branding row summary line does not refresh after Save (#3781) — sheet content is correct, row stays stale. Full screenshots + v…
Brain Voice/Tone CT-BVOICE-10 — authorized real Regenerate, quality PASS (Peyton WP)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Fired real $5 Regenerate on signal-rich QA account. Output specific/grounded (refund policy, ticket-vs-RSVP, Bali audience language, behavioral not adjectival rules) — architecture did not lose touch. Notably the account's PRIOR live voice (still live after Discard) was itself generic ("friendly", adjective-driven) — likely explains the generic-output complaints: stale/pre-architecture voices, not a current generatio…
Brain Voice/Tone post-#3550 re-verify (CT-BVOICE-01..04), Peyton WP
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-BVOICE-01/02/04 PASS; CT-BVOICE-02 confirms #3536 Tweak fix live. CT-BVOICE-03 PARTIAL — exemplar header shows no count for this account; code review confirms this is very likely correct post-#3550 empty-state behavior (fails-closed to only source:'founder'-stamped rows, none exist on this account), not a reopen of #3535 — flagged for a one-line DB check outside this session's read carve-out. CT-BVOICE-04 (guest g…
Brain Voice/Tone (CT-BVOICE-05/06) — second QA account, Peyton Ong
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-BVOICE-06 PASS (all 5 steps). CT-BVOICE-05 PARTIAL: steps 1-6 PASS, step 7 finding — single generate action is gated behind 3 playbook prerequisites not required by the scenario's own precondition (product-call, not filed), plus a duplicated-copy bug in the locked line's text (filed). Step 8 correctly not attempted (nothing to submit).
Brain Voice/Tone re-walk, second QA account — CT-BVOICE-05/06 (closes #3522)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-BVOICE-06 (calibration game) 5/5 PASS. CT-BVOICE-05 (empty-state Generate gather) steps 1-6 PASS, step 7 FAIL — no single always-present "Review & generate" action (AC-VOICE-19); gated behind 3 still-Missing playbook basics instead. Finding A: needs a product call (spec vs. gate). Finding B: copy-duplication bug filed. Step 8 N/A (nothing to accidentally submit).
Brain Voice/Tone (CT-BVOICE-01..04) — baseline walk, Peyton WP
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
3 real bugs found on primary QA account: Tweak silent no-op (#3536), missing exemplar count/rows (#3535), missing Branding-row sub-line (#2957, still open). CT-BVOICE-04 SKIP (no incognito tooling). CT-BVOICE-05/06 BLOCKED (needs second clean account, #3522).
Brain Voice/Tone ("How you sound") click-through — CT-BVOICE-01..06
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial — screenshots captured in-session, not persisted to disk
Peyton WP, prod (PR #3544). Naming check PASS — no "Brand Voice"/"Business Foundation" SKU name shown anywhere in the flow. CT-BVOICE-01/02/03 each PARTIAL FAIL — the Branding row's required sub-line never renders despite a non-empty value; the "Tweak" button is a silent no-op (no panel, no crash, no console error); the "Your voice examples" header shows no count and zero rows. CT-BVOICE-04 SKIP (needs a genuinely si…
On-Brand Salt & Cedar dense reply eval (22 turns)
14/22 replies PASS (64%). Offerings/pricing, honesty-on-unknown, and 3-tier escalation all work correctly. Launch-blocking gap: assistant claims ignorance of its own confirmed location, deposit terms, cancellation policy, and recurring schedule, escalating to itself in the third person instead of answering from the Brain; also lost its own correct answer mid-conversation. Root-cause hypothesis: Brain policy chapter s…
On-Brand Bloom date-fallback re-test (#3995) — CT-DATEFB-00..05
Fix confirmed working end to end on real WhatsApp: with zero live cohorts, On-Brand Support quotes the saved retreat window (14-21 Mar 2026) instead of inventing dates or reciting the season-list trap. Price, deposit, seat-availability honesty, and cancellation policy all held on the same thread. Report: https://mystflo-qa-preview.pages.dev/qa/onbrand-date-fallback-report-2026-08-24.html
Pulse/Usage/Billing IA (#3763) — CT-PUBM-03/04/05 closed via real WhatsApp session
10/11 PASS. Real WhatsApp conversation + accelerated session close unblocked CT-PUBM-03/04/05. Found + resolved a needs-attention naming overlap (not a bug, product note). CT-PUBM-11 still BLOCKED — needs a run-level error, not a conversation escalation.
2 earlier runs for this flow
Pulse/Usage/Billing IA (#3763) — CT-PUBM-10 retry on On-Brand Support
7/11 PASS (CT-PUBM-10 now PASSES on retry via On-Brand Support after Waitlist + Launch #3988 was ruled out). CT-PUBM-03/04/05/11 still BLOCKED — On-Brand Support is WhatsApp-reactive, needs real inbound message for run history.
Pulse/Usage/Billing IA redesign (#3763) frontend/backend sync
CT-PUBM-01/02/06/07/08/09 PASS with real backend cross-checks (dashboard-stats API, raw ai_usage rows byte-matched). Two apparent frontend/backend mismatches root-caused to correct-by-design (wallet-ledger-sourced monthlyCost survives assistant deletion, #3827; KPI-vs-breakdown divergence is the known #3878). CT-PUBM-03/04/05/10/11 BLOCKED: account has zero live assistants; the fixture-generation attempt (activate Wa…
Brain no-charge-scare — copy audit after #3940 (CT-SCARE-00..08)
Prod chunk BrainPage-CNkzuZf9.js confirms #3940 deployed. No wallet-scare copy or leftover AI prices found on live reachable screens (CT-SCARE-00,01,01b,02,08) or in the full deployed bundle text-trace covering CT-SCARE-03..07 (blocked live by this account's near-empty Brain — no research/offerings/voice/playbook). One dead unused constant WRITE_PLAYBOOK_COST:"~$0.50" flagged for cleanup, never rendered. Report: http…
CT-BALI-04 re-run FAIL -- Luma synthesize-profile llm_failed 2/2, blocks CT-07+ (#3966)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
Continuing into CT-BALI-07 found brain_profile empty -- CT-BALI-04's own Result line had never actually been filled this sitting. Re-ran the Luma research pass live to unblock: free identity-peek correctly named The Offline Club Bali (no HQ leak), but the paid confirm (That's me -- run) failed both attempts with brain_read_audit.outcome=llm_failed, model_used=null -- the LLM call never resolved a served model. Crawl …
7 earlier runs for this flow
CT-BALI-05/05b -- IG refuse PASS, screenshot-extract billing-message bug FAIL (#3951)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
CT-BALI-05 PASS (IG URL live-refuse, no charge). CT-BALI-05b FAIL: 2 attempts to extract real IG screenshots both showed a false 'you weren't charged' error while ai_usage actually climbed 1004->1006->1007 (3 real charges). Filed #3951 (P1) with root cause traced to useScreenshotRead.ts. Report: docs/testing/click-through/bali-offline-pilot-findings-2026-08-24-ct-05-05b.md
CT-BALI-06 HQ-vs-chapter fork + wording fix proof (#3920/#3923)
Bare HQ-root paste (theoffline-club.com) correctly forked the identity gate (PR #3920); fallback labels read The part I run / The whole business, not This chapter (PR #3923). Confirm disabled until pick; no paid crawl fired. HTML report with real screenshots: docs/testing/testing-artifacts/bali-offline-pilot-ct-bali-06-report-2026-08-24.html
Bali Offline Club — CT-BALI-14 precondition check (Peyton WP)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Pre-flight precondition check for CT-BALI-14 found the shared Peyton WP tenant reset+reused by the concurrent reactivation QA flow since Bali's 2026-08-20 sitting: brain_profile.business_name=QA Fixture Business, zero TMPL-9P3X/Bali-EVBK01/Bali-EVLK01 rows (Active or Past), only reactivation's own TMPL-RTLK01/TMPL-EVBK01 fixtures present. Did not wipe (would destroy reactivation's live in-flight work) or attempt CT-B…
Bali Offline Club Sitting A — CT-BALI-10..13
CT-BALI-10 PASS (root cause on #3747 corrected to a UX gate, not backend); CT-BALI-11 PASS (new finding: no currency unit, #3753); CT-BALI-12 FAIL (new: second-instance silently discarded, #3752); CT-BALI-13 PASS for hangout run (new: Pulse stale-until-reload, #3754), reading-party run blocked by #3752; CT-BALI-14 inconclusive (viewport); CT-BALI-15 + Sitting B not attempted (need phone / deliberate wipe). Visual rep…
Bali Offline Club CT-BALI-05..09 PASS, CT-BALI-10 BLOCKED on new #3747 (Peyton WP)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-BALI-05/05b/06/07/08/09 all PASS (real Luma + IG paid reads confirmed, FAQs saved via $3 draft path). CT-BALI-10 BLOCKED on new P1 #3747: Try tab stays hard-disabled despite Brain Check saying 'it can run', root-caused to a brain-readiness.ts ready-vs-blockers/attention mismatch. Blocks Activate and the entire downstream tail (Booking, Launch Kit, Pulse, WhatsApp) for the rest of Sitting A and all of Sitting B.
Bali Offline Club CT-BALI-04 re-verification (#3710 fix, Peyton WP)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Re-reproduced the exact #3710 repro path live on prod after PR #3716 merged/deployed; BusinessIdentityGate now mounts and renders visibly (z-index 13500) above the wizard's bare-mode slide-over pane. CT-BALI-04's P0 finding is fixed and verified. Follow-up #3719 (post-confirm completion sheet) filed, not yet re-verified — #3710 stays open.
Bali Offline Club dress rehearsal — Sitting A (CT-BALI-00..04) — direct Chrome, Peyton WP
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-BALI-00..03 PASS; CT-BALI-04 P0 FAIL (Brain research gate bypass, real charge + no UI feedback, #3710 CPO-confirmed policy violation); CT-BALI-05+ and Sitting B blocked on #3710; also found #3711/#3712/#3713
Pulse Activity display — sitting 2 signed-in prod (pulse-activity-display)
CT-PAD-06/07/08 SKIP — Peyton WP wiped today for bali-offline-pilot, 0 assistants/0 runs, plan bars fabricating events. CT-PAD-10 FAIL — /pulse?tab=runs renders Activity correctly (never blank/404/Overview) but URL never rewrites to ?tab=activity per AC-PULSE-24; filed #3929 with root cause + fix in Pulse.tsx resolvePulseActiveTab. Report: docs/testing/testing-artifacts/pulse-activity-display-report-2026-08-23.html
2 earlier runs for this flow
Pulse Activity display + counting — first sitting #3917
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
Preview CT-PAD-01..05,09 PASS. Remaining CT-PAD-06..08,10 signed-in /pulse on prod. Next: /qa run pulse-activity-display
Pulse Activity display + counting — first sitting #3917 (CT-PAD-01..05/09 PASS; 06-08/10 SKIP signed-in)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
Preview + vitest sitting: CT-PAD-01/02/03/04/05/09 PASS. CT-PAD-06/07/08/10 SKIP (need signed-in /pulse). Do not score prod until merge.
Brain Ink & Air — CT-INKAIR-07 Look-paint mechanism investigation (not a scored click-through walk)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-INKAIR-01..06 NOT walked this session (unchanged from 2026-08-18 BLOCKED baseline). Session pivoted to investigating CT-INKAIR-07's underlying mechanism (Look extraction paints Your playbook mono-until-confirmed/paper-after) far outside the plan's original single-account scope: live-tested Research my business / Read your site's style extraction against 2 real QA-account businesses (Female Founders Breakfast, Bali…
1 earlier run for this flow
Brain Ink & Air rollout — Cloud guest (CT-INKAIR-01 deploy gate)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
Guest /brain on production painted the empty workspace (PlaybookPage · 0% · CUSTOMERS HEAR pills), not Screen 1 f34 carousel. Deploy gate BLOCKED. CT-INKAIR-02..06 not walked. Prod bundle still contained carousel copy at run time; #3576 shipped+deployed same-day (post-drain) — re-run needed against current prod. No QA password in this VM.
Reactivation — Chrome Peyton WP (CT-REACT-02/05/08/10/11-Sheet)
CT-REACT-05 PASS (chooser reactivate works). CT-REACT-02 FAIL — new bug #3880 (Pulse Past-tab stopped card missing Reactivate footer, present on Assistants pre-card for same instance). CT-REACT-08 fixture built (EVBK01-B, Team Keepers) but blocked-and-deferred: disconnecting the required WhatsApp app is account-wide and the UI refused since the active EVBK01-A fixture depends on it too. CT-REACT-10 deferred: needs a …
3 earlier runs for this flow
Reactivation sitting 3 — CT-REACT-10/11-Sheet/12 (Chrome, peytonworkprojects@gmail.com)
Sitting 3: CT-REACT-12 hit a real stuck-validation bug filed as #3855 (blocks Waitlist+Launch activation). CT-REACT-11-Sheet blocked on fixture mismatch (EVBK01 uses MystFlo Storage, not Sheets). CT-REACT-10 needs go-ahead (money-failure state on a wallet shared with a concurrent QA session). Report: reactivation-report-2026-08-22-sitting3.html
Reactivation click-through — partial (CT-REACT-03/06 PASS, fix verified live)
CT-REACT-03/06 PASS (direct-bypass vs chooser confirmed on TMPL-EVBK01). Core #3822/#3823 fix verified live: full EVBK01 activation completed end-to-end, previously-broken pipeline now works. 2 new blockers found: TMPL-BRST01 not a real catalog assistant (blocks CT-REACT-02/05), TMPL-EVBK01 is single-instance not multi (blocks CT-REACT-08 as specified, honest-failure fix confirmed on this path too). CT-REACT-01/04/07…
Reactivation — guest walk
Guest Past tab: “No previous assistants.” CT-REACT-01/02/03 need a stopped QA fixture.
Private beta onboarding — CT-PBETA-15 closed PASS (banner Link Google verified live)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Correction: CT-PBETA-15 (banner's own Link Google button, real Google OAuth, single account confirmed via Supabase Auth dashboard) was Peyton's live click today, mis-attributed earlier to CT-PBETA-11 (Sign In modal path, separately also PASS). CT-PBETA-15 flipped to PASS. Peyton approved Option A -- invite now, small batch (3-5 founders). Watch CT-PBETA-07 (exactly-one-email count) on first real sends. Remaining open…
3 earlier runs for this flow
Private-beta onboarding — CredentialRecoveryBanner live walk (#3866/#3869 found+fixed)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Peyton's own live Founder-Action walk (real invitee accounts peytonworkprojects+privatebeta1/2/3@gmail.com) found and fixed 2 new bugs in the net-new CredentialRecoveryBanner (#3860 follow-up to #3836/#3861): banner clipped under sidebar (#3866, fixed), banner Link Google button called wrong function and always failed (#3869, fixed). Both merged+deployed same day. Live-proven PASS today: CT-PBETA-02 (Accept lands sig…
Private-beta onboarding — CT-PBETA-00 gate re-verify (#3836 deploy confirmed)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-PBETA-00 upgraded PARTIAL->PASS: migration 20260822090000 confirmed applied on prod AND notify_auth_signup_welcome() source confirmed rewritten (references email_confirmed_at, not invited_at) via Management API query; deno 6/6, vitest 2/2, EF deployed (405), trigger present. CT-PBETA-05/06 carried forward PASS unchanged. CT-PBETA-01/02/03/04/07/08/09/10/11 remain BLOCKED pending Founder Action (Peyton: Studio crea…
Private-beta onboarding send-now gate
Verdict: FIX FIRST — live invite sitting not run (Send invitation + Accept still owed). CT-PBETA-00 PARTIAL / 05 PASS / 06 PASS; 01-04/07-10 BLOCKED. HTML: docs/testing/testing-artifacts/private-beta-onboarding-report-2026-08-21.html. Next: /qa run private-beta-onboarding after Invite time + Resend id + landing URL.
Pulse/Billing MTD Cost undercount — Peyton WP (CT-PULSEBILL-01..06)
Real bug found (#3827, ~24x MTD cost undercount under prepaid), fixed via PR #3833 (merged+deployed), verified live post-deploy — Pulse/Billing/Usage now all agree with the wallet ledger. Follow-up #3834 filed (non-blocking). One transient error observed, not reproduced, not filed.
Activation Wizard — CT-ACTIVATION-03 (peytontianyan, On-Brand Support) FAIL, dead Complete Setup CTA #3819
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-03 FAIL: Connect row correctly shows Connect (never Verify/Manage); but Complete Setup CTA (detail-activate-cta) is a dead click when Connect step incomplete — zero DOM mutations, zero network calls, no Brain-gate prompt, no Pulse waiting-on-you card. Filed #3819. CT-04 still blocked (real .30 wallet, not safely drainable this session).
2 earlier runs for this flow
Activation Wizard — CT-ACTIVATION-01/02 signed-in Peyton WP (Retreat Launch Kit)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-01 PASS (real $3.00 charged run, AC-BILL-01 satisfied), CT-02 PASS with bug (unresolved [ADDRESS] placeholder leaking into paid output — corroborates existing issue #3396; Brain-readiness 'Not confirmed yet' Try-step warning has no confirm affordance and does not block spend, reproduces tracked blocker, fix PR #3757 still unmerged). CT-03/CT-04 not attempted this sitting (would corrupt shared QA account state othe…
Activation wizard — guest walk (cloud VM)
Guest catalog + Waitlist + Launch pre-detail walked (price/run + Activate visible). **Blocked at Activate:** Get early access waitlist modal, not the wizard. Sign-in had no QA password in this VM. Try/preview sample reply still unwalked. Guest Workshop Connect Apps ✗ is WhatsApp-disconnected — not #3466.
live-email-triggers-postfix-verify — root cause fixed (Resend fallback_value field mismatch), verified live
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Root cause: scripts/create-resend-templates.mjs sent camelCase fallbackValue to Resend's raw REST API (field is snake_case fallback_value); Resend silently dropped it, so no template ever had a working fallback, and mapEventToTemplate never supplies the footer vars (UNSUBSCRIBE_URL etc.) itself -- confirmed via a Peyton-approved platform_errors read: notification_send_failed / 'Missing required variable UNSUBSCRIBE_U…
4 earlier runs for this flow
live-email-triggers-postfix-verify — 2nd post-fix re-verify (PR #3746), still zero Resend sends
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Real $10 top-up on peytonworkprojects@gmail.com (Billing -> Add funds -> Stripe test card) at 2026-08-21T05:00:51Z. Wallet credit confirmed on reload ($223.10 -> $233.10, real ledger row, not optimistic UI). Resend polled 17+ times over ~5min through 05:05:39Z: zero new sends, baseline still stuck at 2026-08-18 08:18:51. Code-audited origin/main: PR #3746's waitUntil fix IS correctly deployed (fireWalletTopupReceipt …
live-email-triggers — post-fix re-verification (incomplete)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
PR #3721 waitUntil fix deployed clean but wallet_topup_receipt still produces zero Resend sends after a real $10 top-up; handed to Cursor Cloud agent for further root-cause (dedup guard / template mapping / call-path candidates).
live-email-triggers — post-persist correction
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-EMAIL-01/04 PASS verdicts were false positives (DB proxy proved code ran, not that Resend sent) — root cause: sendNotificationAsync fire-and-forget with no EdgeRuntime.waitUntil, killed by isolate teardown; fixed in PR #3721 (SHA c655720c3e967cd190596b6f80daceb484fb3a03), deployed via 38-function redeploy 2026-08-20; real-send re-verification still owed, blocked on Chrome extension not connecting.
live-email-triggers
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-EMAIL-01 PASS (topup landed, email-fire unverifiable per plan caveat), CT-EMAIL-02 BLOCKED (fixed threshold tiers cap at $100, QA balance too high to cross cheaply), CT-EMAIL-04 PASS (Retreat Launch Kit activated, assistants.status=active confirmed)
Connection health settle-window gating (#3641, PR #3649) — CT-CONNHEALTH-01..06
01/02/03/05/06 PASS via substitute fixtures (no Google-requiring wizard template exists in catalog); 04 BLOCKED (AI-provider picker not in live UI, AUTO-only AI gate, expected). Zero false Reconnect/Missing-access flashes observed across all runs on the Google-backed Storage row (3x) and On-Brand Support's Connect rows (2x).
Feedback V3 — CT-FB-01 guest PASS; CT-FB-02..07 BLOCKED (no QA sign-in)
Guest page honest (CT-FB-01 PASS at 375 + 1440). Signed-in form + user_feedback not walked — Cloud VM had no QA_USER_PASSWORD / service-role key. No FAILs. No bug filed. This DB row backfilled 2026-08-21 (root-cause fix) — original 2026-08-20 Cloud VM run had no OPS_SUPABASE_* creds so only the markdown snapshot was hand-patched (#3724), which meant every subsequent full-table snapshot regen (DB-sourced) silently dro…
QA bugfix batch #3630-3637 retest (PR #3639) — CT-BUGFIX-01..06
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-BUGFIX-02/03/04/05/06 all PASS clean (guest wizard-draft sweep, Team tabs, erase-data gate, Pulse guest race, Usage-vs-Overview spend total). CT-BUGFIX-01 PASS-narrow + found sibling bug #3643 (preview-cache masks 90-day staleness as Healthy; attempted fix later reverted at PR #3649, real fix #3651). #3644 same-day closed as not-a-bug (wrong account).
Connect funds-first doorway CT-FUNDS-01..06 (PR #3598)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
All 6 scenarios PASS — balance tap opens Top-up pane, chip Change navigates to Billing·Card in-pane with #1419 replace-confirm gate, balance reflects live ~3s post-topup with no reload, never navigates to /billing. Backend curls A-E also PASS (exact $25 delta, one topup row, no double PI, card unchanged).
Our Read deck — CT-OREAD-01..07 against prod (#3562)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Cannot sign in: QA_USER_PASSWORD and SUPABASE_SERVICE_ROLE_KEY not injected as Cursor secrets. App live, Cloudflare bypassed. Add secrets to unblock.
Explore browse + pre-detail modal — CT-EXPBR-01..10
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
7/10 PASS, 2 SKIP (guest signup flows, deferred to avoid disrupting an active session). 2 findings remain: BUG 1 persona-taxonomy doc drift (cosmetic), BUG 2 coming-soon cards show more than the mockup draws (likely intentional, needs mockup/product confirmation). Original BUG 3 resolved as works-as-specified (#3636, test-step error not a product bug).
Waitlist Funnel guest CTA — CT-WLFUNNEL-01
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-WLFUNNEL-01 PASS — the "no dedicated WaitlistModal for guests" finding resolved as stale PRD, not a bug (#3632): guest CTA convergence was deliberate (PR #1124, 2026-06-07), PRD corrected to match. CT-WLFUNNEL-02/03/04 reclassified N/A (blocked under an assumption that no longer holds). 05/06/07 not executed — need the real WhatsApp account, unavailable this session.
Brain shelf + Ink & Air book (PR #3505) — CT-SHELF-01..08
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Peyton WP on PR #3505 preview (not prod). 7/8 PASS: shelf live with two real covers, book open/close, 1024 breakpoint, content matches playbook, Save-a-copy view-only, edit stays on index, design-artifact leak gone (SPA 200 + app 404 body). CT-SHELF-04 SKIP — no incognito. No FAILs.
Brain Check confirm-flip on Connect (#3558 / PR #3568) — CT-BRAINCONFIRM
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Peyton WP prod, build f2688a55 (fix live). Could not observe auto-confirm-flip: business_name is #3593 source-drift (blocking:false, no text edit can clear it); brand_voice is a real content case but unsafe to edit on this shared account (FFB voice on a Bloom instance). CT-BRAINCONFIRM-02 not run. Deploy confirmed; no FAIL evidence.
Event Host Phase 1 catalog — Walk 3 (prod, post-#3482)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
CT-EVHOST-01..04 PASS on prod; 3-card catalog correct, EVLK01 confirmed live beta (PR #3482), no drift.
Brain attention-summary click-through (#3509) — CT-BRAINATT-02 final re-run, root cause resolved
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes — 3 screenshots in-session (Pulse, Connect collapsed, Connect expanded)
Closes out the 2026-08-16 MIXED row below. Root cause was two stacked bugs: **#3540** (write-time verdict floor was fire-and-forget in `brain-profile/index.ts`, could strand a stale verdict — fixed in PR #3541, `await` added at all three call sites, merged+deployed 2026-08-16T02:15Z) and **#3561** (frontend re-derived `blocking` from bucket+status instead of trusting the gate's own flag, mis-flagging a gate-satisfied…
1 earlier run for this flow
Brain attention-summary click-through (#3509) — root-caused a standing Business-name verdict-race bug
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial — in-session Chrome shots not persisted to disk
Testing PR #3509 (`fc521f94`) — the Brain Check summary line for a purely-attention (no hard blockers) state reads `it can {verb} · K to review`, never `before it can {verb}`. **CT-BRAINATT-01** (blocker-branch control) **PASS on both the 2026-08-15 and 2026-08-16 runs** — live line reads `! 0 of 3 ready — 1 needs you before it can take a booking` verbatim, confirming the deployed bundle runs post-#3509 code. **CT-BR…
Signed-in re-verify — wallet / Pulse resume / Event Booking Try
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Peyton WP after #3513/#3514/#3515 on prod. Wallet PASS $147.74→$197.74 live no reload. Pulse resume PASS — Active instantly, one toast, no webhook_url error. Event Booking Try PASS — real preview, no error; yesterday Brain-gate BLOCKED had already cleared. Not Event Host Brain KQA. Still open: Brain #3145, Launch Kit Active-tab empty copy.
1 earlier run for this flow
Signed-in five-flow (wallet / Pulse / Brain / Launch Kit / Event Booking)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
First Peyton WP execution. Wallet FAIL (charge + no reflect → retry double-charged +$100 TEST, account $147.78). Pulse PARTIAL (resume ok, badge stuck Paused, false webhook_url). Brain PARTIAL parked as #3145 paste-organize (product scoping, not a toast). Launch Kit PASS. Event Booking BLOCKED (cross-assistant readiness gate). Fixes on ship-queue: #3513 #3514 #3515. Next prove-walk after drain: wallet → Pulse → Event…
Event Booking + Event Launch Kit — #3516 fix-batch re-verification (7 bugs)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.
Local dev (localhost:5173, CORS-required port) as Peyton WP, re-verifying branch `fix/3516-3517-booking-preview-parity` (5 commits, unshipped). #3517/#3526/#3527/#3528/#3529 fully PASS live. #3516 Personalize-step trigger_type threading PASS live + DB-confirmed; Try-step skipTest UI still BLOCKED by an unrelated pre-existing Brain-readiness "needs review" confirm-UX gap (no tracking issue found — recommend filing). #…
Brain listing layer click-through (#3028) — CT-BLIST-01..09
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial — screenshots captured in-session, not persisted to disk
Peyton WP, prod (PR #3544). 9/9 scenarios PASS — no separate listing section (jump-nav has no extra chip), one-sheet facts display, free assembled default, gather-before-rewrite gate correctly blocks drafting on a zero-facts row, draft-active delete guard, real paid draft ($2.00) → Keep → billing wiring, refine bills at the "from $0.05" micro-op floor (never the full $2.00), a manual edit before Keep correctly flips …
Brain Research ("Read my site") click-through — CT-BRRSCH-01..07
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial — screenshots captured in-session, not persisted to disk
Peyton WP, prod (PR #3544). CT-BRRSCH-02 (URL submit → identity gate self-opens → confirm) PASS; CT-BRRSCH-04 (drafts land on playbook + gap queue) PASS. CT-BRRSCH-03 FAIL — after confirming the paid research bundle, the modal never routed to the live per-source progress checklist; it silently reverted to the blank "Share anything" panel even though the job completed server-side (real $2.50 charge) and the checklist …
Brain FAQ Generation click-through — CT-FAQGEN-01..08
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial — screenshots captured in-session, not persisted to disk
Peyton WP, prod (PR #3544). CT-FAQGEN-07 PASS — the free depth-sheet candidate-picker chip ("+ What's included?") drafts and keeps a real answer with zero AI/price copy anywhere. CT-FAQGEN-01 FAIL — the coverage-dot row never renders (no ●/◐/○, no legend, no "offerings tagged" fragment); a `[useFaqCoverage] coverage fetch failed` console error was observed repeatedly, the row fails closed with no error or empty state…
FAQ per persona re-walk (CT-FAQPP-01..03)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
Cloud /qa run: preview live at cursor-faq-per-persona-4330; deploy gate UNREACHABLE — no Peyton WP session in cloud Chrome. CT-FAQPP-01..03 not graded.
Event Host Brain KQA (CT-EVHB-01..06)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial — in-session shots not persisted
Claude in Chrome as Peyton WP at 1440 and 375. Chip + Bali Offline Club example PASS. Chip order FLAG (#3494 not on origin). FAQ candidate chips FAIL — picker only surfaces Main (Retreat host); Event-host night-loop corpus never reaches a secondary persona.
Ledger proof — Chrome Peyton WP (CT-LEDGER-02)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial — 500w not 375; shots not in this checkout
Claude in Chrome as Peyton WP. All four steps PASS: signed-in catalog + prices, Retreat Launch Kit Overview $3.00/run (no Complete Setup), Pulse Overview + Active + Past at ~500 then 1440, no crash, nothing activated or spent. Cold-load flake: first /assistants paint briefly showed On-Brand Support unactivated while a second tab showed Running as FFB Support; reload fixed. Console AbortError + account-status/profile …
1 earlier run for this flow
Ledger proof — Cloud guest (CT-LEDGER-01)
Walked CT-LEDGER-01 guest on platform.mystflo.com. Catalog cards + prices; header Sign in + Get early access (no founder name). Waitlist + Launch Overview showed $0.20/run and Overview / Personalize / Connect / Try. Complete Setup opened Get early access (name / email / WhatsApp), not the wizard. Closed without submitting. CT-LEDGER-02 not started: Cloud header was guest, not Peyton WP; did not invent a password.
Wallet top-up — guest walk
Guest `/billing` is $0.00 + “Sign in to view.” No Add funds. Did not enter a card (would not risk a live charge). No `account_wallet` side-effect possible.
Pulse monitoring — guest walk
Guest empty Overview/Active/Past rendered (0 runs, All healthy, “Nothing needs you”). That is not CT-PULSE-01 — no post-card to reconcile. Tabs are Overview · Active · Past · Runs (draft still says Waiting). Pause/resume not attempted.
Launch Kit post-activation — guest walk
No activated WMKT01/RTLK01. Guest only opened RTLK01 pre-detail Connect (Apps ✓ · Wallet ✗). Not the post-activation modal.
Brain playbook LOOK f9/f33 — guest walk
Guest `/brain` is empty intake (“Research my business”). Deploy gate (white book button on a populated playbook) not reachable. Does **not** prove LOOK is missing from Pages — #3459 merged 2026-08-14; re-run as Peyton WP.
Connect apps — #3446 live walk + RTBK01/WSBK01 activate attempt
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial
Walked wizard Connect and post-tab on Peyton WP against prod. Hypothesized "missing Payment row" is wrong — seeded roles are `chat(whatsapp)` + `storage(mystflo_storage)` only. **Failed:** RTBK01 and WSBK01 stay Apps ✗ / Try disabled because `allRolesPassed` looked up a `connected_apps` row that internal MystFlo Storage never has (#3466). **Passed:** row inventory matches seed; Wallet ✓ is the card, not a Payment rol…
Pulse status / attention / timestamps lock (PR #3416)
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.yes
First-executed Claude-in-Chrome on Peyton WP. **No FAILs.** Passed: attention>0 inbox (CT-PSA-01 A), legend 1:1, Created + Status since, quiet Runs dots, Active + Past tabs. **Unverified residuals (not FAILs):** health-only empty, both-queues-empty, wallet_empty Top up, Failed/Stopped + Load older, manual-paused Breakdown row.
R7 — FFB live WhatsApp conversation, post-#3358/#3409/#3410
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.no — phone shots exist with Peyton, not in this report
Activation blocker from Round 6 is gone (#3151 fixed) — first real conversation ran. 6-turn live WhatsApp thread, Peyton's phone → FFB number. **Passed:** 4/6 #3358 reply-style guards (promised-artifact + escalation-dedupe hold); billing (`chat_sessions.billing_status=charged`); Pulse `total_runs` 3→4 on session close (#3242). **Failed:** price-verbatim — "low four figures (Euro)" said verbatim despite a seeded €999 …
Brain FAQ journey — 16 scenarios × 3 businesses (FFB, Uglywoo, Saltline)
Executed the FAQ journey plan against prod. **Passed:** bracket-fabrication guard (PR #3254) confirmed live across 5 trips. **Failed then fixed same day:** paid-draft billing bug (#3280, PR #3285).
R6 — 5-business On-Brand Support accuracy eval
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.partial — 8 shots, biz #1 P1/P2 only
Biz #1 (FFB) Brain fully seeded (8/8 required fields) but activation was a dead end — Try tab disabled, Complete Setup loops back to Connect, all 3 Brain-Check gate items fail through confirmed bugs (Business hours save silently lost, Contact details deep-link goes to a panel with no matching field, Brand voice paid $5.00 + saved but readiness flag never clears). 6 defects filed (#3145–#3151, two P0). Playbook for al…
R5 — Ambre Studio (carry-over) — sheets behind the doors
The `(look)` sheet leaks the same unclamped wall the row does; the `(sound)` door reads a different, empty column. Corrected two Round 4 findings in place.
4 earlier runs for this flow
R4 — Ambre Studio (carry-over) — landing surfaces, post-#2988
Branding row overruns design rule V-12 by ~47×, and only on the Instagram-only path. First round with real screenshots.
R3 — `@ambrestudio` Instagram-only
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.no — text only
Clears the §8.5 hard pass bar on all 6 markers. Extraction fidelity high; defects are all post-extraction. Predates the screenshot contract — text + JSON only.
R2 — `@uglywoobali` Instagram-only
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.no — text only
First genuinely clean baseline. Exact follower/post counts (174/650/114). Predates the screenshot contract — text + JSON only.
R1 — Female Founders Breakfast, website + Instagram
No screenshot was saved for this walk. The Chrome session kept the shots and they never landed in git.no — text only
7/7 event-schedule rows matched verbatim. **Ran against a contaminated crawl pool** (reset-script bug, fixed in #2998) — claim grading valid, "clean baseline" framing is not. Predates the screenshot contract — text + JSON only.