Sculptape — Brand Database (BRAND.md)
The single source of truth for every durable fact about this brand. Any note David sends
(inventory, suppliers, accounts, contacts, decisions) is filed HERE the same turn it arrives —
before any context compaction. Memory files only POINT here. Never store secrets (tokens, keys,
passwords): reference the env file that holds them. Keep entries dated. Hub renders this file.
0 · Identity & entities
| Field | Value |
| Legal entity / owner | TO FILL |
| Key people & roles | TO FILL |
| Support email / phone | TO FILL |
| Brand relationship | TO FILL (own brand / client / partner) |
1 · Products & SKUs
| Code | Product | Size / format | UPC | Formula / spec source | Supplier |
| TO FILL |
2 · Inventory & supply chain
| SKU | Units on hand | Location / 3PL | Lot / mfg date | Shelf life | COGS (landed) | MOQ / reorder lead time | Updated |
| ST-01…ST-06 (pack of 10 tapes) | TO FILL | Lansil (3PL) | $1.09 COGS + $0.20 sea freight = $1.29 landed (David, Costs tab 2026-09-19) | 2026-09-19 |
Notes: fulfillment per order (David 2026-09-19): pick/pack $1.20 · packaging $0.30 · postage $3.78 flat (no per-extra-pack step) · payment 2.9% + $0.30 · shipping-protection claims 10% of bump revenue. Source of truth: tools/proforma/assumptions/sculptape.json; edit on the Analytics → Costs tab.
3 · Offer & pricing (locked)
- Ladder (David 2026-09-22, house-wide BOGO formula): Buy One $24.99 (1 Pack, the zone's base id) · Buy 2 + Get 1 FREE = 3 packs at 2× = $49.98 (pkgs 64 / 66 / 68 / 70 / 72 / 74 for Silhouette / Waistline / Lift / Neckline / Profile / Sleeve) · Buy 3 + Get 2 FREE = 5 packs at 3× = $74.97 (65 / 67 / 69 / 71 / 73 / 75); compare-at $24.99 × packs → Save 33 / 40%; Free Bonus Pack unlocks at ≥ $60 (tier 3). The old 2 Packs $37.50 / 3 Packs $44.99 packages stay in campaign 1838 for order history only (still on the checkout allow-list).
?ob=c's second option = Buy One at the regular price; the one-time trial lane is unlinked. - Try Before You Buy keep charge = DAY 7 (David 2026-09-17: "its pay on Day 7 not 14 for sculptape"). Both lanes: sub $0 → $19.99 on day 7 then every 30 days (pkgs 5/11/17/23/29/35); one-time $0 → $24.99 on day 7 (pkgs 6/12/18/24/30/36); $6.99 shipping at order. 29next
interval_count7 on all twelve (aligned 2026-09-17); copy on the PDP offer box, checkout CAT sub-lines, Terms §3.2 ("seventh (7th) day"), Refunds;check_compliance.pyasserts the day-7 wording. - TO FILL — retail, compare-at, tiers, subscription, guarantee, OTO chain, shipping policy, launch code.
4 · Accounts & IDs (no secrets)
| System | ID / handle | Notes |
| Domain(s) / DNS | TO FILL | |
| Netlify site (id · team) | PDPs/checkout sculptape c9d5d07c-037b-43e5-99fa-1a70dcedbe05 · EcomScalers (teo-h3uxfci) — sculptape.com<br>Prelanders sculptape-prelanders 616390cc-ab0d-4973-81d2-5887565914de · EcomScalers (moved 2026-09-10)<br>Launch board sculptape-launch-board 928f3649-9a05-484e-aaa9-8e3d05cb56cb · Riptide (admin-ragrhxe) | House rule: brand sites live on EcomScalers |
| 29next store / campaign | TO FILL | |
| Meta BM · ad account · pixel | Ad account from 2026-09-20: act_268669251366940 "RS – Spare" · BM AMV LLC 113824746806442 (David). Until then: act_243417265464399 "Liaison – B" · BM DeltaSift 244352822784717 — its Sculptape campaigns are PAUSED, history stays there. Pixel 1065595576190917 "Sculptape": owned by DeltaSift / Liaison–B, shared to AMV LLC and assigned to RS – Spare (verified on the pixel's shared_accounts edge). Page "Carol Anne Foster" 927552797118674 is owned by AMV LLC. | CAPI token → env file only. Proforma + Analytics pull BOTH accounts (meta.accounts); campaign_match "Sculptape" keeps the shared account's other advertisers out. |
| GA4 property · measurement id · GTM | TO FILL | |
| Klaviyo org | TO FILL | |
| adscalerads client · Airtable base | TO FILL | |
| Cookie / event prefix | TO FILL |
- Meta ad account (since 2026-09-20):
act_268669251366940"RS – Spare" — BM AMV LLC113824746806442, a shared/spare account (David: cheaper CPMs than Liaison–Bact_243417265464399, where the 2026-09-17→19 campaigns ran; those are paused). Pixel1065595576190917(DeltaSift-owned, dataset "Sculptape") is shared to it. Launch Page from this account: Smooth and Clear104489138535397(AMV; Carol Anne Foster927552797118674is AMV too; Fifty and Glowing104493539168657sits in the Seasonal Products BM → needs a share before it can promote). The analytics pipeline pulls both accounts (multi-account sync, 2026-09-20).
5 · Compliance ceiling
TO FILL — category rules, banned words file, reviewer rulings, disclaimer form.
Competitor comparability (David 2026-09-05, HARD): anything Sculptape is ranked against must be
INSTANT, WEARABLE, NON-SURGICAL and NON-TOPICAL. "A lotion doesn't count, or any plastic surgery."
Out for good: firming creams (NIVEA Q10, StriVectin), clinics/surgery (CoolSculpting, abdominoplasty,
mastopexy, platysmaplasty, rhytidectomy, brachioplasty), injectables, devices (NuFACE), overnight
patches (Frownies — "drop frownies"). Shapewear/adhesive bras STAY as the what-you-own-today foil
("keep the shapewear ones": SKIMS, SPANX, NuBra). Competitor copy: accurate, fair, live-verified
prices, no fabricated defects — Tonee sells PRE-CUT clear shaped tapes ($41/10 sale, $49 sticker,
"can be trimmed and shaped"), never call it "a roll".
6 · Decisions & taste log (dated, David's verdicts = ground truth)
- 2026-09-29 — The swipe boards are published for the team under the Launch Center, no password (David: "is there a way to publish this boards on our launch center so my video editors can see them and prelander designers can see them", then "no password, put it under launch-center"). https://launch-center-hq.netlify.app/swipes/ lists every board in
tools/brand-hub/brands.jsonads.swipes; each board is a read-only copy of the J&T board (pictures, the ad text, the Facebook Page, the page it sends people to, its Ad Library link; search and filters; the kept flag from David's newest verdict export). Rebuilt bytools/swipe-site/build.pyon everytools/deploy-launch-center.sh --build. Unlisted and noindex, but anyone with the link can open it. Verdicts stay on the J&T board on this Mac. - 2026-09-29 — Launch structure + ad codes (HARD, house-wide; David: "when launching the same ad for differnet adsets and different landing pages, we need to keep the name of the ad the same, maybe AD###" · "moving forward we need to have 1 prelander per campaign, each ad should have its own adset, and the adsets should be ABO of $50 per adset"). One prelander per campaign, one ad per ad set, ABO $50/day per ad set, the ad and its ad set named with the ad's code. Launch with
tools/ads_launch.py(dry run, then--goon David's word per launch); results by code withtools/ad_codes.py report --brand sculptape --pages; kill withad_codes.py kill --brand sculptape AD###. Live ads from before the rule keep their names (learning-phase rule) and are mapped to codes inbriefs/sculptape/ad-codes.json. - 2026-09-26 — VERDICT (every brand): the mobile checkout has no hero photo and no caption under it. David, on a Charm mobile screenshot: "on mobile checkout, please remove photo, and the caption", then "make that the case across all checkouts". At ≤767px the page runs header → secure bar → steps → product title → stars → promo box → offer. Desktop keeps the photo.
- 2026-09-22 — Offer box v2 + BOGO formula + Try First "Buy One" (David, house-wide: "now do it across all brands for the main offer and ?ob=c"). 2 Packs + 1 Free $37.50 / 3 Packs + 2 Free $44.99 → Buy 2 + Get 1 FREE $49.98 (3 packs) / Buy 3 + Get 2 FREE $74.97 (5 packs) — 12 new packages 64–75 via
build_29next.py --create(BOGO_TIERS), map regenerated. Deltas to know: 3 packs cost $5 more than the old 3-pack; the top tier goes 3 → 5 packs for $30 more. The stacked-row box (tools/offer_box_v2.py, copy inproducts/sculptape/offer_box.json: Buy One / Free shipping · Buy 2 + Get 1 FREE / Save 33% + Free Shipping! / MOST POPULAR · Buy 3 + Get 2 FREE / Save 40% + Free Shipping! / BEST DEAL / ✓ Free Bonus Pack) replaces the tiles on all six zones;?ob=c's second option = Buy One (pk1, "Buy One — $24.99", sticky follows). Imagery gap: CMS slots 14 / 15 still show the 2-pack / 3-pack renders — a 3-pack and a 5-pack shot per zone are needed (David). Built + QA'd in the worktree (6 PDPs × 2 viewports 0 failures, gate PASS); not deployed — Sculptape deploys only from /dev. - 2026-09-21 — Launch default (house-wide, David: "moving forward on launches - No audience network & no advantage+ on placements"): MANUAL placements — Facebook, Instagram, Messenger; no Audience Network, no Advantage+ placements. Set in the adscaler launch route (
createAdSet→publisher_platforms) andtools/jt-launch-post.py; SKILL §VIII "Launch defaults". In force since 2026-09-21 18:06 PDT (adscaler deploymentdpl_G9KuEjpXX47BB234MXCrxWmEmqpZ) — every ad set created after that carries manual placements. The campaigns live today were launched BEFORE the rule and still run Advantage+ placements incl. Audience Network — untouched unless David says to edit them (a targeting edit on a live ad set resets learning). - 2026-09-21 — Regular-flow test vs
?ob=c(David: "launch the same charm and sculptape campaigns that we currently have enabled, but not run them to the ?ob=c because we fear it may be jacking up our CPMs. Lets try the regular flow" · "Lets set those to $50/day also"). Twins of the two live RS – Spare campaigns with the plain PDP URL (no?ob=c→ the PDP's default offer box instead of the Try-Before-You-Buy box) at $50/day each: images120249814450970266(4 ad sets, CBO $50 = 4 × $12.50, the same 59 ads/copy) + videos120249814458050266($50 on the single ad set, the three videos reused by id). Same page, pixel, targeting. The?ob=coriginals (120249800312530266$100 ·120249799953330266$100) were left RUNNING — David did not ask to pause them, so both lanes run side by side and CPM reads like-for-like (same creatives, account, days). Charm twin:briefs/charm/BRAND.md§6. - 2026-09-20 — Relaunch into RS – Spare = ONE headline + ONE primary text per page, one ad set per page, one CBO campaign; image ads → the PDP
?ob=c, NOT the prelanders. David: "i want all of the ads to have 1 headlines and 1 post text instead of the multiple, that way i can put all of the images into 1 adset per prelander, and then i can make another campaign with 1 adset, with 3 video ads with the respective pdp?ob=c" → "instead of building that, we can do a one-off launch where i pick the headlines/posttext i want per page and you can just launch them in the new account". His picks — Thighs H2/P2 · Neck H1/P2 · Face H1/P1 · Tummy H3/P1: Thighs & Booty H "Beauty Editor Ranks 5 Cellulite Fixes" / P "Most women think you need compression to hide cellulite. You don't. A beauty editor reviewed 5 popular solutio…" · Neck & Jawline H "The #1 Neck Tightener for 2026" / P "Sagging neck skin isn't a cream problem—it's a gravity problem. You need to pull, not push. The #1 ranked neck…" · Face Lift H "Top 5 At-Home Face Lifts of 2026" / P ""We tested 5 products on whether the face actually lifts today, and whether it survives someone sitting across…" · Tummy Line H "The Tummy Trick Editors Now Use" / P "A 2026 review just ranked the top 5 ways to flatten the tummy line. The criteria: Does it work instantly? Does…" (full lines intools/jt/one-offs/sculptape-relaunch-2026-09-20/sculptape-relaunch-plan.json). Structure: campaign A = 4 ad sets (Thighs & Booty 29 · Neck & Jawline 13 · Face Lift 12 · Tummy Line 5 = 59 image ads), CBO $100/day (4 × $25); campaign B = 1 ad set, 3 videos (#511 thighs → the-silhouette · #547 neck → the-neckline · #541 tummy → the-waistline)?ob=c, $100/day on the ad set (a one-ad-set campaign is not CBO — same spend). The first campaign A went to the Top-5 prelanders; David ten minutes later: "Sorry, i didnt want the images going to prelanders, i wanted them going to the pdp?ob=c for the image ads. My mistake" → "Just pause that prelander campaign and launch another one" — paused, rebuilt tosculptape.com/products/{the-silhouette,the-neckline,the-profile,the-waistline}/?ob=c(page → PDP via the pages-manifestdestination; Face → the-profile). Second time image ads moved prelander → PDP (the 2026-09-19 split was the first): Sculptape image ads default to the PDP?ob=c. Page = Smooth and Clear (104489138535397, AMV LLC — a page the RS – Spare system user can promote from; Carol Anne Foster is AMV too; Fifty and Glowing lives in the Seasonal Products BM and needs a share). Old Liaison–B campaigns paused by David himself. Recipe + the Meta API budget rule: SKILL §VIII "Relaunching a brand into a NEW ad account". - 2026-09-19 — Question posts launched as POSTS (David: "those are a different type of ad, it's like a post, not an image ad"; "Carol Anne Foster, $5/day boost post"). The format runs as a persona-Page photo post with the card as the photo and NO link, promoted for post engagement (campaign
120249494369340521"Sculptape (ST-01) — Question posts · Carol Anne Foster", OUTCOME_ENGAGEMENT, one ad set per post at $5/day, POST_ENGAGEMENT / ON_POST, US · women 35–65 hard-targeted — Advantage+ audience caps the age control at 25, so it's off here). The three kept r8 cards (bride → Silhouette, neck → Neckline, weight → Waistline) are up PAUSED; the link and the proof live in the pinned UPDATE comment (PDP?ob=c), which the Page has to post by hand until the system user getspages_manage_posts+pages_manage_engagement(comment pack + before/after images sent to David; dark post ids increative/sculptape-ads/launch-posts-2026-09-19-1734.json). The second "customer" comment: from a REAL profile, never a persona. Tool:tools/jt-launch-post.py. Meta lessons: an engagement ad set needsdestination_type=ON_POSTor Meta demands a website URL; campaigns without CBO needis_adset_budget_sharing_enabled. - 2026-09-17 — LAUNCH INCIDENT (no spend): the first Sculptape launch from the J&T bar went to Meta with the Netlify staging origin as the landing page. David at ~18:0x: "Cancel the ads that's launching right now, urgent, we used the wrong prelander url, it's still on netlify." State found: campaign
120249458996640521"Sculptape (ST-01) — All" (Liaison – B, act_243417265464399, $300/day, created 18:00:29) ACTIVE with zero ad sets, zero ads, zero spend — the launch job died after the campaign shell. Paused at the account level at ~18:2x; watched 5 more minutes, nothing appeared. Root cause: the 62 saved ads (and 79 copy rows) carriedhttps://sculptape-prelanders.netlify.app/…aslanding_page_urlfrom when they were saved; the prelanders' real domain is shapeandsculptreviews.com (David: "shapeandsculptreviews.com is the domain for the netlify prelanders"; the pages-manifest had been moved to it in commit 4f595b35b at 18:04, four minutes after the launch). Fixed:POST /api/saved-ads/relink-landingon adscalerads rewrote saved_ads + copy_generations for clients 25/24 (62 ads, 79 rows) to the new origin; the video-inbox pages.json rebuilt; Copy Studio extractorPBmoved. Standing law: the Netlify origin is staging only — never a landing page in an ad; the launch bar and the saved ads must carry the brand domain, and the saved ads' landing URL must be re-checked against the pages-manifest before every launch. The paused campaign shell is left in Ads Manager for David to delete. - 2026-09-17 — CHECKOUT DEAD 2026-09-15 05:25 → 2026-09-17 (David: "big error… the checkout says the total price is $68.29… the total order summary is always 68.29… the auto-complete on the address isn't working… the country box shouldn't be there right"). Root cause: the
/next-api/*→ campaigns.apps.29next.com proxy lived only innetlify.toml, andtools/deploy-sculptape.sh(staged-dir deploys since the 2026-09-12 hygiene pass) runs the CLI from the repo root, where no toml is read — every [[redirects]]/[[headers]] rule vanished, the proxy answered Netlify's HTML 404, the campaign-cart SDK never initialised ([CampaignStore] Campaign load failed: 404), so the summary kept its baked $68.29, no package loaded, the country select stayed raw ('Select Country', empty) and Google Places (booted by the SDK) never started. NOT the 29next Allowed-Domains gate (the API answered 200 direct). Fix:products/sculptape/_redirects+_headersgenerated from the toml (tools/netlify_toml_to_files.py), deployed; proxy 403-JSON again, legal 301s back. Verified live:?ob=c→ TBYB → cart line pkg 5 $0.00 (recurring $19.99 / 14 d), shipping $6.99, Shipping Protection $9.99 pre-checked → total $16.98; address autocomplete returns Places rows; country defaults to US with the 10-countryshowCountrieslist fromcheckout/sculptape/config.js(Charm/CellScience show 244 — the field is by design; US-only would be David's call). Guards: the deploy script refuses a stage without_redirectsand fails if the proxy is 404 after deploy. Law in CLAUDE.md. - 2026-09-17 — Prelander links →
?ob=c(David, house-wide). All six Mirror Test pages land on their zone PDP's Try Before You Buy box: forwarder stampsob=c(inbound?ob=awins for QA), static hrefs carry&ob=c; donor-fixed sobuild_top5.pyreproduces it; the stale 'sculptape-microinfusion' forwarder comment corrected. Param forwarding ON. Domain shapeandsculptreviews.com confirmed (root 302 → thigh-smoothing); pages-manifest live_urls moved off sculptape-prelanders.netlify.app to the apex. Deployed viatools/deploy-st-prelanders.sh. Then David: "shapeandsculptreviews.com is the domain please make sure the prelanders are that domain" → host-level 301 (_redirects), Copy Studio extractor + JSONs, adscalerads video-inbox pages, netlify.toml and the factory skill all moved to the domain; no Meta ad linked to the netlify.app host. - 2026-09-17 — 45 refine tags for COPY, not images (David: "the images are great, but the copy is not good, please classify all the copy here and then give me 5 copy options per that I can pick for you to re-render"; his one note, on r3-legs-wikihow: "Headline doesn't make any sense: '20 seconds, then look.' What does it even mean? Let's be a little bit more deliberate and plain."). Classification of the 45 lines: cryptic time cues ("Twenty seconds. Then look."), insider metaphors ("The chin-up you do in every photo.", "The lift you trust up top. Now for the rest.", "Under the hairline, yes."), statistics without the product ("Nine in ten women have it…"), contrasts without the noun ("Creams ask you to believe…"), confessions without the product, dares — one fault throughout: the line never says what the product is. LAW: a static's line names the noun (a clear body tape) and says plainly what it does. Picks: r3 → option 1, r4 → 2, r5 → 1, r6 → 2, r7 inferred (same option as the matching composition, else 1) — all 45 rendered as v2 on the same cards; four corrected to v3 in David's own words ("nobody knows what jowl is" → the loose skin along the jaw; "Half your face without wrinkles, with a clear tape. The other half... not yet."; "Clear tape lifts your butt. Clear and smooth."; "No cellulite or wrinkles. A clear body tape."); "rest is great". No "jowl" in copy. Five plain options per card written on the Truth Gate (
ads/research/body-tape/copy-refine-2026-09-17.md) and put on the cards as pick chips; the pick swaps only the copy on v2 (edit pass on the same image). Also converted to refine: r3-face-split, r3-neck-photo, r3-neck-profile, r5-wiki-legs-2 (earlier "perfect" / "amazing" — the image stays, the copy changes). - 2026-09-17 — Round 8 verdict (3 keep / 3 kill): bride (ST-A09), neck (ST-A10), weight (ST-A06) kept; swim, stylist, postpartum killed. Then David sent the rest of the reference thread: "thats the part i didnt screenshot. They also add b/a images, and theres the second comment too. lets add those to the 3 ones i kept" → the three keeps now carry the author's own stacked before/after above the link in her UPDATE ("edit: mine came. here's my first day with it, the second photo is with it on") and a second commenter's testimonial with her side-by-side before/after (Jenna Marchetti 8 months / Colleen Ashby 5 months / Tasha Reid 7 months — fictional, "tape on in the second one", no medical claims). Pairs built with the before/after skill; the bride pair was refused as underwear from behind and passed as bike shorts pulled up. The thread's three beats are now in the framework, the adscalerads twin and Template G (G3). Round 7 also judged: 25 keep / 5 kill (no notes yet recorded here — see the verdict file).
- 2026-09-17 — New ad format in the framework library: the organic question post + UPDATE comment (David: "Please add this type of text ad to the framework. This is the organic ugc question and comment response. So we can make this part of the generated playbook." — reference: the "korean peptide serum instead of botox" post, 1.1K likes · 736 comments · 44 shares). Image framework
ugc-text-status-question(coloured Facebook text card with a lowercase crowd question, rendered with exact words), copy framework 'Organic Question + UPDATE Comment' in adscalerads, Template G in the DTC templates. Round 8 = six Sculptape question posts (creative/sculptape-ads/r8/), persona pages with fictional names, the third party always unnamed, each with its pinned UPDATE comment on TRUE claims only: bride/bridal stylist (ST-A09), swim coach (ST-A03), weight-loss nurse (ST-A06), mother of the bride / makeup artist (ST-A10), celebrity stylist (ST-A01), swim-class mom (ST-A13). The comment's link = the campaign page; price $24.99; 30-Day Money-Back Guarantee. Board = 191 statics, rounds [1–8]. Then David: "i think it has to be setup in this format, the hook has to be strong" (the reference's first line) → the setup is FIXED: every card opens "i saw an interview where celebrities said they use a", only the category / zone / fix / result change; the six cards were rewritten before any verdict (none judged) and the framework, the adscalerads twin and Template G carry the fixed hook. - 2026-09-16 (late) — Refine is a verdict now (David: "we need a refine button on the cards, keep kill refine, that way we can have a state on the card and you can update it with the new version of the photo, so we can check between the previous version and this version"). The J&T board has ✓ KEEP · ↻ REFINE · ✗ KILL; a refine keeps the card with a note on what to change,
tools/jt-refine.pyrenders the next version of the SAME card (<id>.v2.jpg, the previous file untouched), the card shows v1 · v2 pills to flip between them plus the note that produced v2, and reads as a new version awaiting his verdict. First refine answered:r7-wiki-thinv2 (the stray face inset removed). Kills and keeps never get a new version under the same id — only refines do. - 2026-09-16 (late) — VERDICT on round 6 (16 keep / 13 kill) → round 7 = every re-render in its own round. All 12 wikiHow cartoons kept (the lane is 24 of 25 over three rounds), the three necks kept, the jeans tummy kept; every butt / legs photo killed, the leggings tummy killed, the four faces killed — three with the note "Needs to be closer to the hairline/temple area to be exact. It can't be that close to the eyebrow." Timing truth: David judged r6 at 22:16–22:23 and the in-place butt-law fix landed at 22:31, so the 8 butt/legs kills were the old thigh-band renders and the 7 cartoon keeps the pre-fix versions. David: "Can you put anything re-rendered in another round, because I marked them as killed, so I don't know what's changed" — "So all the re-rendered in Round 7." → STANDING RULE: never overwrite a judged card; every re-render goes into the next round under a new id. Round 6 restored byte-for-byte to the versions he judged (commit 40931d190); the 15 cheek-law renders carried into round 7 as new cards (
carried_fromon the row). Face placement law: the strip starts AT THE HAIRLINE above and behind the temple — where the hair begins, level with the top of the ear — never beside the eyebrow, never on the temple skin near the eye; referencecreative/sculptape/identity/images/w10-worn-face-hairline.png= the r4 profile keep edited (two from-scratch attempts put it on the temple skin and then turned it into a headband; editing a keep worked first time). Round 7 (sculptape-shots-r7.json, 30 shots,creative/sculptape-ads/r7/): the 15 carried cheek-law renders (six from-behind photos, two from-behind splits, seven cartoons), four faces at the hairline, the leggings tummy on the jeans keep, eight more wikiHow (thin-and-still-have-it, tags still on, the cage comes off, in the shower, instant tuck, six zones, own hand, on the towel), two more necks. Board = 185 statics, rounds [1–7]. - 2026-09-16 (later) — THE BUTT LAW, with David's example photo (a round-6 from-behind shot, the strip on the right cheek): "So all of the butt ones, the tape goes from UNDER the underwear to the middle of their buttcheek. Not around their thighs, not on their side thigh. This is a good example except it needs to be wider." And: "If you want to show them wearing shorts or biker shorts, make it so they are pulling it up to reveal it underneath." → On every butt shot the tape is a WIDE clear panel (about 4 inches wide) centred on the cheek, its top under the underwear / bikini / shorts edge and its lower part visible down to the MIDDLE of the cheek — never on the thigh, never on the side of the thigh; the r5–r6 "short wide band on the back of the thigh" stays for THIGH shots only. With shorts or biker shorts, her hand pulls the leg of the shorts up to reveal the tape underneath on the cheek. Placement references rendered from the note:
creative/sculptape/identity/images/w9-worn-butt-cheek.png(bikini-cut underwear, the panel from under the edge to mid-cheek beside a really dimpled left cheek) andw9-worn-butt-shorts.png(biker shorts pulled up high to reveal it) — the first try from a biker-shorts base put it back on the thigh because the shorts cover the cheek; the reference has to expose the cheek. Round 6 fixed IN PLACE before judging: 15 shots re-rendered under the same ids (the six from-behind photos, the two from-behind splits, seven cartoons: bikini, mirror, swimsuit, 16 hours, postpartum, runner, creams). Same file paths — hard-reload the board to see them. - 2026-09-16 — VERDICT on round 5 (J&T board, 8 keep / 23 kill) → round 6 = the wikiHow lane doubled, real swipe keeps as problem anchors, the tape smaller and wider, faces temple-to-crown and happier. Keeps: r5-wiki-butt-2 "This is exactly it." · r5-wiki-legs-2 "This is amazing." · r5-wiki-20sec, r5-wiki-dress, r5-wiki-face-2, r5-wiki-tummy-2 (the cartoon lane is 12 of 13 over two rounds) · r5-ba-neck-event (the one two-day split that passed) · r5-legs-scam. Kill notes, verbatim: r5-ba-legs-back "You have to make the tape smaller. Should be peeking underneath the pants" · r5-butt-beach "Can make the tapes smaller and wider" · r5-butt-mirror "Make the non-taped leg have more cellulite, and the right leg with the tape on it be even smoother." · r5-butt-20sec "Make the left leg have more cellulite, the one that doesn't have tape on it." · r5-butt-hand "Need to show the other buttcheek with more cellulite, and no need to show the hand quarter turned. full behind" · r5-ba-tummy-leggings "This tummy doesn't show a flap being folded up. Nor does the before show the flap of the stomach going OVER the waist line" · r5-ba-face-wedding "The before photo: her face sagging even more" · r5-ba-neck-front "Make the after photo of her face look a bit more lifted and younger." · r5-face-half-2 "gotta make her happier looking" · r5-ba-face-profile "And the tape needs to run from her temple to the back of her head to her hair." · r5-face-mirror-2 "The face tape is like a facelift. It pulls from the temples up to the crown of the back of the head, the other side of the temple." · r5-wiki-neck-2 "Make the before photo here of a saggy neck." · 11 kills without a note (the other two-day splits, legs-bikeshorts, legs-hemcheck, legs-shorts, neck-chinup-2). The laws, updated: the tape on legs and butt is a WIDE, SHORT band — as wide as the thigh, about 2 inches tall — its top under the hem and only its bottom edge peeking 1–2 inches out, smaller than a hand, never a tall rectangle; the untaped side heavy, REAL cellulite; the taped side perfectly smooth; shot straight from behind, no hands, no quarter turn. Tummy: the before flap hangs OVER the waistline, the after shows it FOLDED UP. Face: the tape runs from the TEMPLE up and back to the CROWN into the hair, like a facelift string; the woman looks HAPPY and the after visibly younger. Neck: the before is a really saggy neck. Method change: the model under-renders the problem from words alone, so the problem side now anchors on David's REAL swipe keeps (Lymphlite's dimpled cheeks 1863802908340038 and thighs 974290758719386, the bikini-from-behind 2392325097944511, CryoContour's loose stomach over the waistband 4592280747765103, the flap 1873158820262331, the saggy-neck profile 2197621031041163, the lifted face 27700103696251974, the crepey arms 982991064327873); two new placement references rendered from the notes (
w8-worn-thigh-band.png,w8-worn-face-crown.png); cartoons carry their own DNA (the photo DNA pulled two into photographs). → Round 6 (sculptape-shots-r6.json, 29 shots,creative/sculptape-ads/r6/): 12 wikiHow (neck with a saggy before, runner, bikini from behind, one-piece vs two-piece, face temple-to-crown, arms with the creams, swimsuit vs shapewear, the 16-hour timeline, postpartum with the kids, the bodycon Day 1 / Day 9, the mirror check, creams vs 0:20), 6 single-frame from behind, 8 two-day splits, 3 faces/necks. Board = 155 statics, rounds [1–6]. 13 re-renders across three passes (cartoons as photos, label leaks, a serif headline, an odd face inset, a collage, both sides taped, a tall band, a pregnant-looking pooch, gloss on a before neck, one safety refusal in underwear → bike shorts). - 2026-09-16 — VERDICT on round 4 (J&T board, 21 keep / 8 kill) → round 5 = the before/after skill on the splits + "tape easier to see, cellulite more". The placement law landed: every wikiHow cartoon (6/6), every face (4/4), every neck (3/3), 5 of 7 legs, both tummies with the tuck, the arm split — kept, no notes. Kill notes, verbatim: r4-butt-mirror + r3-butt-hand "Okay this is good, make the tape easier to see, and make the other butt cheek sag and cellulite more." · r4-butt-20sec "Okay this is good, make the tape easier to see, and make the non-tape leg cellulite more" · r4-butt-hand "Okay this is good, and make the non-tape leg cellulite more" · r4-tummy-jeans "great now use our before after skill on this, remember to make the model slightly before and after more, can't have the exact same everything, looks too AI" · r4-butt-beach, r4-legs-bikeshorts, r4-legs-shorts, r4-tummy-beach: kill, no note. New law: a split is two SEPARATE photos of the same woman weeks apart — different day, room, light, hair, clothes, makeup, the SAME framing of the affected area (the before-after-generator skill's three rules); identical-everything-but-the-skin reads as AI. And the tape must be easy to see (one clear glossy rectangle, crisp edges, a bright sheen), the untaped side heavy and obvious in sidelight. → Round 5 (
ads/research/body-tape/shot-lists/sculptape-shots-r5.json, 31 shots,creative/sculptape-ads/r5/): 13 splits through the skill (tools/angle-statics-gen.pyshotmode: "ba"— before pass, then the after with the before as identity reference on a different day + the placement anchors as product refs, two 540×1350 portraits side by side, the hook line typeset by PIL in Helvetica Neue Bold as band / caption / comment card, so the frames are free to differ and the text is never a render fault); 6 single-frame kills re-shot with the notes; 6 more wikiHow on new hooks; 3 faces/necks and 2 legs on new hooks. Anchors = the placement references + David's r3/r4 keeps. Board = 126 statics, rounds [1–5]. Render QA: 7 of 31 re-rendered — the two-pass neck afters put the tape on the throat (fixed by naming the bare zones: front, throat, side, under the jaw), one after pass returned a labelled BEFORE/AFTER collage inside the frame (the after prompt now forbids collages), the neck cartoon drew the panel on the side of the neck, the dress cartoon on the thigh, two leg shots too faint; the beach tummy after was refused twice as a bikini and passed as swim bottoms + a tankini at a lake. - 2026-09-16 — VERDICT on round 3 (J&T board, 4 keep / 20 kill, 1 unjudged) → the TAPE-PLACEMENT LAW of round 4. Keeps: r3-face-split "This is perfect." · r3-neck-photo "This is perfect." · r3-neck-profile "Perfect." · r3-legs-wikihow "This is great, but the left leg needs more sunlight. More of these". Kill notes, verbatim: r3-butt-mirror / r3-butt-beach / r3-legs-bikeshorts / r3-legs-normal "The main feedback is that the tape needs to be higher up, tucked underneath the shorts, and it should be peeking out about 3 inches below the shorts, pulling the skin up. One side will look smooth, and the other leg will look like cellulite." · r3-butt-track "This is perfect, but the right leg from behind should be slightly pulled up to expose the tape, and then more sunlight on the non-taped leg" · r3-butt-20sec "This is a great example of the tape being tucked underneath the shorts, so just make the other leg, the non-taped leg, have more sidelight." · r3-legs-comment "Again, the right-side shorts should be pulled up, exposing the tape underneath, and the other leg should have more sunlight." · r3-legs-dance "The dress should be slightly raised on the taped leg and should be exposing the tape from underneath. The non-taped leg should have more silhouette and more wrinkles." · r3-legs-16hrs "As with the other ones, the tape needs to be peeking underneath. It needs to be hidden underneath the underwear, and then it should be peeking out about a couple of inches. The other leg should have more cellulite, and the taped leg should have none." · r3-legs-shorts "More cellulite on the non-taped leg" · r3-tummy-tuck "This is almost there, but it needs to show, on the 'before,' a stomach that's going over the leggings or the pants. The 'after' should show the flap folded into the stomach with the tape over it." · r3-tummy-jeans "Same feedback as the other stomach one. The stomach flap should be sitting over the jeans, and then the after should be tucked up into the belly button with the tape over it." · r3-tummy-rolled "Yes, the stomach needs to be tucked inside the dress, and tape should be showing, peeking right underneath it." · r3-neck-portrait "Tape should be pulled from the back of the neck, not in the front." · r3-face-half "I think the tape should be a little more visible on the face to show that the left side, or in this case, the right side, is pulled up." · r3-face-mirror "The plastic tape actually goes from the temple behind the hair, not going down to the jaw." · r3-face-profile "Tape doesn't work like that. The tape pulls from the temples back into the hair." · r3-arms-split "The tape should be less visible." · r3-legs-track, r3-tummy-beach: kill, no note · r3-butt-hand: unjudged. THE LAW: legs/butt — the tape sits HIGH, its top hidden under the hem (shorts / underwear / bikini / dress), its lower edge peeking out about 3 inches below as one clean rectangle, pulling the skin UP; the garment on the taped side pulled up slightly to expose it; the taped leg has NO cellulite, the untaped leg MORE, lit by sidelight / sunlight. Tummy — before = the flap hanging OVER the waistband; after = the flap folded UP into the belly button with the tape over it; under a dress = the stomach tucked inside the dress, the tape peeking right underneath. Neck — pulled from the BACK of the neck, nothing in front. Face — the tape runs from the TEMPLE back into the HAIR, slightly visible, never down to the jaw. Arms — the tape barely visible. → Round 4 (
ads/research/body-tape/shot-lists/sculptape-shots-r4.json, 29 shots,creative/sculptape-ads/r4/): six wikiHow cartoons ("More of these" — legs with sunlight on the untaped leg, butt, tummy, neck, face, arms), a v3 of every noted kill with its note applied, three necks and four faces modelled on the perfect keeps, one arm. New placement references rendered FROM the notes and used as anchors:creative/sculptape/identity/images/w7-worn-thigh-hem.png(under the hem, peeking ~3 in, hem lifted),w7-worn-tummy-tucked.png(flap over the waistband → tucked to the navel),w7-worn-face-temple.png(temple back into the hair); the neck and face anchors are the r3 keeps David called perfect. Board = 95 statics, rounds [1–4]. 10 of 29 re-rendered after the contact-sheet pass (two wikiHow shots came out as photos; one arm split stacked into two frames; one dark patch for the panel; the rest needed the tape or the cellulite pushed harder). - 2026-09-16 — VERDICT on the native rounds 1+2 (J&T board, 14 keep / 12 kill) → the LAW of round 3: the problem AND the result on the SAME body in ONE frame. Every kill was a prop or still-life shot (the makeup bag, the tote, the gym bag, the quad grid, the zones diagram, the "twenty seconds" clock card, the "believe" card, the rank card, the swimsuit-on-a-hanger, the pool edge, the "weeks" calendar). Every keep is a body — and every keep note says the same thing, verbatim: a03-h4-rolled "This needs to have more of the stomach flap rolled up. Right now, it doesn't look like anything." · a03-h5-comment "One side should be a clean leg like this one. The other one should have wrinkles and stuff on it." · a08-spell-normal "One side should say 'normal' (with all the cellulite kind of spelling normal), and the other side should have the tape pulling it up, looking smooth." · a08-h1-nine + a01-h5-native "The right leg should have less cellulite, and the left leg should have more." · a06-h4-comment "Should show more of the belly being tucked in." · a11-wikihow "This should have one side with the leg looking cellulite and one with none, with the tape on it." · a10-h3-comment "It should show the before and after of the neck being loose, and then this is the after, where the neck is tight." · a03-h4-dance "Tape needs to be on the bottom. That should show one of the legs being smooth and the other one having cellulite above the knee." · a10-portrait "This tape should be on the back of the neck, smoothing it out, and then before, with all the hanging, sagging skin." · a04-skin + the 3 Video Studio cuts: keep, no note. Then: "please generate more. and also some faces too. and some shots of the butt from behind, one cheek being lifted the other sagging." → Round 3 (
ads/research/body-tape/shot-lists/sculptape-shots-r3.json, 25 shots,creative/sculptape-ads/r3/): a v2 of every noted keep with its note applied (one leg dimpled / the other smooth under the panel; the belly loose / then tucked; the neck hanging / then held by the band at the NAPE); 5 butt-from-behind shots (one cheek lifted by the panel, the other sagging — mirror, beach, track, own hand, 0:20); 4 face shots on the string-lift kit (ST-P5: half-face lifted vs drooping, split selfies, side-profile jowl, mirror tuck — anchorw4-worn-face-stringlift.png); 1 arm split. Style block carries the law verbatim. Note for the record: earlier angle risk notes said "no before/after body imagery" — David's keeps and notes explicitly demand the with/without on one body; David's verdict is ground truth. - 2026-09-16 — VERDICT on rounds 1+2: "Why are all of these so bad? They are all beige. We don't want branding in these, we want the best conversion." → STANDING LAW for every Sculptape ad (and every brand): statics carry NO brand DNA, NO ivory/champagne palette, NO carton, NO wordmark, NO editorial serif. They look like the kept swipes — phone photos of real bodies in real rooms, with/without splits, small white caption boxes in bold sans, the clear panel on skin as the only product. The brand lives on the landing page. Both rounds re-rendered native the same day (same ids / angles / hooks / frameworks,
--commentin the shot lists); the beige renders were discarded. Generator style block (tools/angle-statics-gen.pyshot-liststyle.dna) now says exactly this. - 2026-09-16 — Round 2 statics generated (David: "okay I've approved a lot of source and feedback ads, now let's generate, use the framework library and the approved ones"). 28 statics on the J&T board (
creative/sculptape-ads/r2/), built from David's 23 swipe keeps × the image framework library × the angle bank: the keeps clustered on Lymphlite's "it was never fat" (→ ST-A08 ×4), Tonee's creams / shapewear / no-surgery / in-seconds lines (→ ST-A03 ×3, ST-A07 ×2, ST-A11 ×4, ST-A04 ×3), CryoContour's "lost the weight, the skin didn't" (→ ST-A06 ×3), the bikini / one-piece story pages (→ ST-A13 ×2, ST-A02 ×3) and the neck / jaw keeps (→ ST-A10 ×4). Frameworks: personal-confession-portrait, frustration-layering-still-life, problem-resolution-split, bait-curiosity-single-subject, quad-grid-photo, handwriting-on-skin-ugc, ill-wikihow-cartoon, spots-spell-a-word (dimples spell NORMAL). Campaigns = page codes (ST-P1 thighs · ST-P2 tummy · ST-P4 neck · ST-P6 arms). Shot listads/research/body-tape/shot-lists/sculptape-shots-r2.json; generatortools/angle-statics-gen.py(house Gemini call, locked renders as anchors, hook line typeset in-image). Two re-rendered for faults (a13 towel: headline duplicated on the bag with a typo; a03 dance: floating carton). Price in-image = $24.99 (the live PDP), not the packet's $29.99. The angle board is STILL unclicked — this is a pre-verdict round like r1; a KILL there retires its shots. - 2026-09-16 — Ads are judged on the J&T board only (David: "retire the hub statics board, judge on J&T only"). Hub board
sculptape--top5-statics-r1deleted, console askst-top5-statics-r1withdrawn, replaced byst-r1-judge-jtpointing atlocalhost:8766/dashboard/#sculptape. The Launch Center keeps angles + hooks (sculptape--angles); the J&T board keeps ads. Standing rule for every brand. - 2026-09-16 — Ads playbook started (Angle-Studio-first — SKILL §VIII). Sculptape wired into the J&T ad pipeline: board
sculptape→creative/sculptape-ads/(adscalerads clients 25 full / 24 synth) + swipe librarysculptape-swipe→creative/sculptape-swipe/(GROUP = angle category, chips = the citing angle cards;tools/build-angle-swipe-manifest.py). Static-image Ads Library sweep r1–r3 (David: "do facebook ads library swipes of static image ads also"): 24 + 32 + 32 queries → 1,069 creatives (873 statics + 196 video posters) after the kill-on-sight purge. Niche truth: the tape brands run zero active statics (video only); the statics live on the shapewear giants (SKIMS, SPANX, Maidenform, Shapellx, Leonisa, Harper Wilde) and the pain-vernacular story pages (Cellulite Reset, Lymphlite, Miami MD, 45+ And Thriving, SayBeam, Comfort Curve, Sarah Mitchell, Prime Time Women)., both registered intools/brand-hub/brands.json. Round 1 on the board = the 13 Top-5-thigh statics (campaignST-P1, group = angle) + Raj's 3 Video Studio cuts. Campaigns = page codes from the pages-manifest (/api/brand-pages). Gate: the 13 angles are still CANDIDATE — nothing routes until the angle board is verdicted; the hub boardsculptape--top5-statics-r1duplicates the J&T round-1 surface (judge onlocalhost:8766/dashboard/#sculptape). - 2026-09-16 — Top-5 thigh prelander statics r1 (David: "develop static ads for the prelander") — the first statics built FROM the Angle Studio:
creative/sculptape/statics/build_top5_statics.py→ 13 statics (1080×1350), each one angle's hook in one format across the four angles the studio routes to the ranking page (ST-A09 ×4 incl. the ranked-list card · ST-A03 ×3 · ST-A11 ×3 · ST-A01 ×3), packetsbriefs/sculptape/angles/<ID>/static.md+adcopy.mdwritten with--forcebecause the angle board is still unverdicted (pre-verdict round, said so on the board). House nano-banana call with the locked panel + carton renders as anchors; hook line typeset in-image; three re-rendered for typography faults. Boardsculptape--top5-statics-r1(verdict, pick 6) on the Launch Center; console askst-top5-statics-r1; manifest carriesangle+hookper image (CL-20). Target/top5-thigh-smoothing/utm_campaign=st01_top5. - 2026-09-15 — ANGLE STUDIO round 1 boarded (
briefs/sculptape/angles/): 13 angles inseeds.md(ST-A01 makeup for your body · A02 shorts for the first time in years · A03 invisible where shapewear can't go · A04 sixteen hours, waterproof · A05 the wedding is in nine days · A06 after the weight came off, the skin didn't · A07 the $8,000 fix and the $29.99 one · A08 cellulite isn't a weight problem · A09 boob tape for the rest of you · A10 the neck gives it away · A11 twenty seconds in the mirror · A12 try it first · A13 postpartum and still in the swimsuit), 65 hooks inhooks.md, every seed on TRUE ids only (the banned pseudo-mechanism layer + p6/p13/p14/p15/i7 recorded under KILLED). Evidence: first Meta Ad Library sweep for the category (ads/research/body-tape/: body / tummy / face / bust tape keywords, 276 creatives) + the Deep Research 08 VoC step (config.json → their_words). Boardsculptape--angleson the Launch Center; console askst-angle-board-r1; the Angle Studio station now sits in every workspace's left nav. CL-20 in force: every Sculptape ad, prelander, script and shot list is written FROM a GO angle's packet and names its ID. - 2026-09-14 — Guarantee renamed 30-Day Money-Back Guarantee (was the 60-Day Love-the-Mirror Guarantee, with 90-day template residue on the checkout and a 120-day line in a top-5 page). David's cross-brand law (2026-09-14: every brand uses a 30-Day Money-Back Guarantee; HairScience + WAM excepted). Flipped at the source —
build_site_pages.pyGUARANTEE_DAYS 60 → 30 + naming,build_pdps.py,build_checkout.py(90d.webp → 30d.webp shared 56x56 twin, group-1-30.png shield repainted, subtext, 'within 30 days'),pages_content.py(refunds/terms windows),check_compliance.py(guarantee=30d),build_thumbs234.py/build_thumb6.pyre-rendered,prelanders/sculptape/build_top5.py(six zone pages rebuilt). Shipped-HTML residue 0 old windows; compliance PASS; site + prelanders deployed through the gated scripts and verified live. - 2026-09-02 — BRAND.md created.
- 2026-09-05 — Top-5 prelander rivals = comparable-only (see §5). Approved roster, per page (rank 2→5): Thighs = Tonee Legs & Booty ($41–49/10) · Skinnies Instant Thigh Lifts ($25) · Hollywood Fashion Secrets Body Contour Tape ($13–25) · SKIMS Sculpting Short. Tummy = Tonee Tummy · Skinnies Tummy Tuck ($25) · BOOMBA Body Tape ($15–25) · SKIMS Bodysuit. Bust = Brassybra ($38/3 sets) · Cakes Body · Booby Tape · NuBra. Neck = Tonee Neck Tape ($45–53, 2 tensions, 10 wears) · Secret Lift Pro Neck & Jaw ($18/12) · Skinnies Neck Lift ($30/15 pr) · Art Harding's ($20). Face = Secret Lift Pro Face & Eye ($18) · Art Harding's · Bring It Up ($6–22) · Skinnies Face ($25–30). Arms = Tonee Arm ($41–49) · Skinnies Arm Lifts ($25, 5 pr) · BOOMBA · SPANX Arm Tights. Face page's third score axis became Comfort (every kit is reversible, so Reversible said nothing). Stills: generated trade-dress studio stills (nano-banana-pro) in `creative/sculptape/references/ cmp-assets/branded/` — tonee-panels, tonee-neck, skinnies-{thigh,tummy,arm,neck,face}, hfs-body-contour, boomba-body-tape, brassybra, secret-lift-pro, art-hardings, bring-it-up.
- 2026-09-06 — Editorial voice = a BEAUTY editor, never a "body editor" (David: "it's not BODY EDITOR, it's a Beauty Editor, change across"). Applies to every Sculptape prelander subhead/meta and any future author line.
- 2026-09-06 — The five ranked things are PRODUCTS, not "options" (David: "call them products as that's what they are and easier to understand"); breadcrumb section is Beauty (was Body & Beauty). H1s keep the listicle form "The 5 Best Ways To…" unless David says otherwise.
- 2026-09-27 — Radio-row package selector on all six PDPs + the brand checkout (David, house-wide: "can you also make all the other pdps package selectors look like that? Same for the checkout"): 3 Packs (Most Popular) · 5 Packs (Best Value) · 1 Pack, per-pack price against the $24.99 pack + Save pill (the 1-pack carries none — no discount), then "or" + the Try Before You Buy row ($0 today · $19.99 if you keep it · ships every 30 days · cancel anytime) linking the zone's TBYB subscribe lane. The checkout keeps the TBYB lanes link-only (six zones on one checkout, like PeptideScience's two products) — no trial row; the zone thumbnail rides each row on wide screens. Same turn: Total Order Summary under the Complete Purchase button, the 2-line 30-Day Money-Back Guarantee badge card (accent
--callout), no price on the sticky bar — every PDP.
- 2026-09-27 — The sticky bar names the package, not the deal (David: "the name on the bottom bar on mobile and desktop doesn't match the name of the package"): the bar title is the selected row's name — "The Waistline — 5 Packs" (was "The Waistline — Buy 3 + Get 2 FREE") — on mobile and desktop (one element,
#stickyBundleTitle); the trial row keeps its own title. LIVE the same evening. Follow-up 2026-09-30: the title shown before any row is tapped still read the deal ("The Lift — Buy 2 + Get 1 FREE") — the 09-27 fix renamed only the per-row titles. The initial title is now the default row's name, "The Lift — 3 Packs" etc. on all six PDPs (products/sculptape/build_pdps.py), LIVE.
- 2026-09-28 — No price on the Add To Cart button (David: "please remove the price from the add to cart buttons across all pdps and brands"): the button says "Get The Waistline" etc. (was "Get The Waistline — $49.98"); the trial row's button says "Try It for Free"; the ?ob=c Buy One button says "Buy One". The price lives in the rows, the sticky bar's price stays hidden (09-27). Applied in
products/sculptape/build_pdps.py+offer_box.json; deployed the same day.
7 · Asset map
TO FILL — raw vault, official logos/wordmarks, packaging files, photography, UGC, boards, Visual DNA.
8 · Live surfaces
- Swipe boards for the team (read-only, no password, since 2026-09-29): https://launch-center-hq.netlify.app/swipes/sculptape-swipe/ (the competitor ads, for the video editors) · all brands: https://launch-center-hq.netlify.app/swipes/
- Hook Studio (hooks judged on their own screen — 👍 / 👎, keys G · B · N): https://copy-sheet-hq.netlify.app/sculptape/?tab=hooks · 65 hooks · same verdict log as the angle board · taste digest
briefs/sculptape/angles/TASTE.mdafter the first pass - Round-1 ads (13 Top-5 thigh statics + 3 Video Studio cuts) — judge on the J&T board: http://localhost:8766/dashboard/#sculptape · images
creative/sculptape-ads/r1/(sourcecreative/sculptape/statics/r1/). The hub boardsculptape--top5-statics-r1was RETIRED 2026-09-16 (David: "retire the hub statics board, judge on J&T only"). - Prelanders ST-P1..P6 — The Mirror Test, DOMAIN = https://shapeandsculptreviews.com (David 2026-09-17: "shapeandsculptreviews.com is the domain"): /top5-thigh-smoothing/ (root 302s here) · /top5-tummy-smoothing/ · /top5-bust-lift/ · /top5-neck-tightening/ · /top5-face-lifting/ · /top5-arm-firming/ — Netlify site
sculptape-prelanders(616390cc…, EcomScalers), deploy ONLY viatools/deploy-st-prelanders.sh;sculptape-prelanders.netlify.app/301s to the domain (_redirects). Every reference uses the domain: pages-manifest live_url, Copy Studio (tools/copy-studio/extract.pyPB +tools/brand-hub/copy/sculptape/.json), adscalerads video-inbox pages, the factory skill. No Meta ad links to either host as of 2026-09-17 (checked act_243417265464399 + the LL1 accounts). - Angle Studio board (round 1, awaiting verdicts): https://copy-sheet-hq.netlify.app/sculptape/?tab=angles · bank
briefs/sculptape/angles/
TO FILL — PDP, home, checkout, hub workspace, boards, console lane, docs pages.
- Copy sheet (2026-10-28, David: every headline and primary-text idea judged on one spreadsheet-style grid, on every brand — copy law CL-23): https://copy-sheet-hq.netlify.app/sculptape/ — 65 rows (65 hook), nothing judged yet;
python3 tools/angle-studio/copysheet.py build --brand sculptape --auto·verdictsfolds the 👍 / 👎 intobriefs/sculptape/copy/. · since 2026-10-30 the brand's page on Copy Sheet HQ carries every tab: angles · hooks · hook buckets · ad headlines · ad primary text · email subject lines · the truth gate · every verdict list (the Launch Center's grids and card views retired behind redirects)
9 · Open questions / asks
TO FILL
10 · Change log
- 2026-11-02 — order line under the checkout button (David, from the NailScience board, "for all checkouts … under the checkout button is the order summary, but it would just say the Item/ea price" → lane "B"): a light-grey "Your order" card right under Complete Purchase with the cart's item and its price each, from
tools/checkout_order_line.py(reads the 29next cart; hidden until the cart holds a product; add-ons such as Shipping Protection left out; no MutationObserver, no motion). Card only, straight from the module with the rows' units (a full rebuild carries undeployed selector + summary changes). Committed, NOT deployed:tools/deploy-sculptape.shruns only from the /dev checkout (shared/cms) — deploy from /dev after the P step. The Item / Amount table and the cart-mode panel stay. - 2026-10-01 — Economics gate (playbook Phase 1c) at a $60 CPA: KILL; break-even ROAS on a first ladder order 1.44 (David: "put the breakeven ROAS as another box … AVERAGE ORDER PRICE / LEFT AFTER COSTS"). Subscription at $19.99: $11.54 left per pack; LTV after costs $21.88 / $36.51 / $51.16 (pessimistic / industry / optimistic), so even optimistic retention never pays $60 back. One-time: the average order ($35.83 on an even mix; 1 ladder order so far) leaves $24.84, so 59% of buyers would have to reorder after every order. It would pass at a CPA under $32.82. Break-even ROAS = $35.83 ÷ $24.84; below it, a first order loses money after ads. https://launch-center-hq.netlify.app/sculptape/economics/
- 2026-10-01 — The house retention assumption is David's cohort table (David: "yes, use my table as the house assumption"): day-7 keep 70% → 90.24%, churn by cycle 28.8 / 21.6 / 18% → 25.21 / 26.70 / 27.67 / 47.12 / 39.01%, then 12.6% → 33.14% a month (the average of his month 2–6 decays); the VIP schedule is unchanged; the previous values stay in
tools/proforma/assumptions/sculptape.jsonunderprevious_retention. The Launch Center's Analytics tab (Forecast + LTV) now runs on it; the Forecast also reads the schedule past its three sliders. Model: https://claude.ai/artifact/Doex81SMjd8a7Zqqp9zxfF. The hub Costs tab's saved keep rate (70%) and fallback churn (18%) were updated to 90.24% / 33.14% so they no longer override it. - 2026-10-01 — Subscriber cohort model + the first trial charges (David shared a cohort table — "use these … cohort averages and decays"): https://claude.ai/artifact/Doex81SMjd8a7Zqqp9zxfF models one new subscriber on that curve with Sculptape's $19.99 renewal ($11.54 contribution a paid month after the $1.09 pack, $5.48 fulfilment, fees and 5% refunds). Our subscribers so far (29next, synced 2026-10-01): 3 trials since Sep 19 — both subscribe-lane trials (subs 397, 463) were charged the first $19.99 pack (orders 103763, 103862), then refunded and cancelled ("not satisfied with the product"); the one-time trial (sub 499) paid $24.99 (order 103895, authorized). So 0 of 2 subscribe-lane first charges kept.
- 2026-09-29 — Swipe boards published at https://launch-center-hq.netlify.app/swipes/ (sculptape-swipe); the Launch Center's shared tools and each brand's Images & Creative tab link them.
- 2026-09-29 — Ad codes backfilled read-only from
act_268669251366940(RS – Spare): 62 codes inbriefs/sculptape/ad-codes.json. No ad renamed. - 2026-09-26 — Mobile checkout without the hero photo + caption (the verdict in §6): the rule comes from the shared template stylesheet through
build_checkout.py, which now versions the combined stylesheet link by content hash (combined.min.css?v=…) —/checkout/sculptape/*is cached 7 days, so a CSS change used to take up to a week to reach a returning visitor. Deployed from the main checkout (tools/deploy-sculptape.sh). - 2026-09-22 — BOGO ladder + offer box v2 + Buy One built (NOT deployed):
build_pdps.py(tiles / ATC / sticky / cart defaults / BUNDLES from the BOGO package names, planone→ pk1 in PLAN_PKG, plan-awaresyncSticky+ newsyncCta,offer_box_v2.applylast so the CMS-slot<img>tags survive),build_checkout.py(cards $49.98 / $74.97, strikes $74.97 / $124.95, "Save 33%", PKG_INFO + TIERMAP labels, order-summary BOGO rows, allow-list keeps the old 2/3-pack ids),tracking/st-cart.js(ZONES.idsreplaces the base+N arithmetic, TIERS 49.98 / 74.97 with retail 74.97 / 124.95, 'one' 24.99 in TBYB_ATC_VALUE,?ob=c/plan=links pass through the drawer), collection band;tools/qa/sculptape_cart_matrix.pyre-specced (D3 $99.96, D5 {1, 65}, C1 "5 Packs", C2 pkg 65, C2z pkg 67, O2 Buy One, C3 sub lane + Buy One in the normal selector, X1 pkg 64). Deploy = David from /dev:git pull→bash tools/deploy-sculptape.sh→python3 tools/qa/sculptape_cart_matrix.py --quick+python3 tools/qa/offer_box_check.py --brand sculptape --live. - 2026-09-21 — Angle Studio: Hook Studio screen live (
sculptape--hooks, 65 hooks, batch 1); hooks judged there feed TASTE.md and the next batch (angles.py hooks --batch 2) lands flagged NEW. - 2026-09-21 — Regular-flow twins launched (through adscaler):
120249814450970266"Sculptape (ST-01) — Images → PDP (regular) · single copy" — ad sets120249814451310266(Beauty Editor Ranks 5 Ce… ×29) /120249814453970266(The #1 Neck Tightener fo… ×13) /120249814455190266(Top 5 At-Home Face Lifts… ×12) /120249814456620266(The Tummy Trick Editors … ×5), campaign daily_budget 5000;120249814458050266"Sculptape (ST-01) — Videos → PDP (regular)" — ad set120249814458220266, ads120249814458940266(Beauty Editor Ranks 5 Cellulite [#511]),120249814458900266(The #1 Neck Tightener for 2026 [#547]),120249814458880266(The Tummy Trick Editors Now Use [#541]), 5000 on the ad set. Landing =sculptape.com/products/{the-silhouette,the-neckline,the-profile,the-waistline}/(no parameter). Verified on Meta right after launch: both campaigns ACTIVE, CBO 5000 / ad-set 5000, all 62 ads on the plain PDP URL with the single title/body, page 104489138535397, pixel PURCHASE, US 18–65 Advantage+ (a share of the new ads sat in PENDING_REVIEW at read time — normal for fresh ads). Runner/payloads/results:tools/jt/one-offs/regular-flow-test-2026-09-21/.?ob=coriginals untouched. - 2026-09-20 — RELAUNCHED INTO RS – SPARE (
act_268669251366940) — David's one-off (§6). Video campaign120249799953330266"Sculptape (ST-01) — Videos → PDP (ob=c)" ACTIVE: ad set120249799953430266$100/day, ads120249799954060266(#511 → the-silhouette) ·120249799954040266(#547 → the-neckline) ·120249799954020266(#541 → the-waistline), all?ob=c, page 104489138535397, pixel 1065595576190917 PURCHASE, videos 1511812850973759 / 1444223284260490 / 2083180365723552 — verified on Meta (3 ads ACTIVE/IN_PROCESS, links right, no Netlify). Prelander image campaign120249799959310266"Sculptape (ST-01) — Top-5 prelanders · single copy" (4 ad sets, 59 ads, CBO $100) launched 22:14 → PAUSED 22:19 at David's ask (wrong destination — images belong on the PDP; left paused, not deleted). PDP image campaign120249800312530266"Sculptape (ST-01) — Images → PDP (ob=c) · single copy" ACTIVE 23:04 (held by the watcher until Meta's hourly API budget had headroom at 23:03): CBO $100/day (campaign daily_budget 10000), ad sets Thighs & Booty → the-silhouette120249800313360266(29 ads) · Neck & Jawline → the-neckline120249800315420266(13) · Face Lift → the-profile120249800316400266(12) · Tummy Line → the-waistline120249800317840266(5) = 59 ads ACTIVE, one title + one body each, pixel 1065595576190917 PURCHASE, US 18–65 Advantage+ audience, page 104489138535397. Verified on Meta 23:12 (one creative per ad set + a video ad read directly): page 104489138535397, link = the respective PDP?ob=c, headline + primary text = the picks, image/video attached. (Reading creatives back needscurl -g— acreative{…}field expansion is otherwise globbed into two bogus URLs and comes back empty, which looks like a wrong launch.) Meta API budget finding: the adscaler app is on Marketing-API development access — 3 video uploads + the 59-ad launch leftx-business-use-case-usage total_timeat 79% for the hour and reads then failed "User request limit reached"; the launch route has no backoff +maxDuration 60→ rule in SKILL §VIII: read the header before any big launch. Files:tools/jt/one-offs/sculptape-relaunch-2026-09-20/(plan, runner, watcher, payloads, results). Records: memoryjt_ad_launch_architecture+sculptape_launch_status, vault "J&T Ad Pipeline". - 2026-09-19 — VIP rebill was DEAD for ~30 h (fixed). David: "the Sculpt Tape VIP upsell isn't attached to a recurring rebill of $9.99". Cause: my
build_29next.py --alignrun for the day-7 change (2026-09-18 16:23 UTC) PATCHed the six zone VIP packages (51–56) withoutis_recurring, and 29next flips the flag to FALSE on such a PATCH — interval day/30 and price_recurring $9.99 stayed, so nothing looked wrong, but every VIP accept after that (test 103138; REAL order 103202, 2026-09-19 08:15) booked the $12.49 line with NO $9.99/30d subscription. Package 44 (membership, retired) kept its flag because its spec sends it. Fix: PATCH 51–56 withis_recurring: true(client campaign API confirms recurring True on all six); the align routine now sends and comparesis_recurringon every package. Proven by a sandbox order accepting VIP (webkit run st-tbyb-20260919-2240, 22/22: VIP subscription day/30 $9.99 + lane subscription day/7 $19.99). Side fix in the same pass: the relay's test-order gate (shared/functions/order-webhook.mjsstep 3.6) used to RETURN before steps 8b/8c, so QA trial subscriptions kept the $6.99 shipping on their renewal while real ones were clean; it now only skips the Purchase send (step 8) and the subscription housekeeping runs for every order. Open for David: TWO real customers accepted VIP inside the dead window and have no $9.99/30d subscription — order 103202 (user 1321, 2026-09-19 08:15, TBYB-subscribe lane, has the day/7 $19.99 subscription 397) and order 103269 (user 1388, 2026-09-19 17:54, full-price lane $47.49 + VIP $12.49, no subscription at all). Both paid with PayPal (vault on file). The dashboard cannot create a subscription and the admin API's create-subscription takes a card token only, butPOST /api/admin/orders/{number}/add-line-items/with{product_id: <VARIANT id, e.g. 137>, quantity: 1, price: "0.00", is_upsell: true, subscription: {interval: day, interval_count: 30, price: "9.99"}}charges $0 today against the order's payment method and creates (or merges into) the customer's day/30 subscription — proven on sandbox order 103307 (the line merged into its day/30 sub 433). Awaiting David's go before running it on real customers. Note: the upsell chain is the Silhouette's for every zone (/checkout/→/products/the-silhouette/special-offer/vip/), so both customers accepted a Silhouette VIP pack even though they bought Waistline / Neckline. - 2026-09-19 — Real costs landed via the Analytics Costs tab (David typed them the same evening the tab shipped): COGS $1.09/pack, sea freight $0.20/pack, pick/pack $1.20, packaging $0.30, postage $3.78 flat per order. Filed into §2 and
tools/proforma/assumptions/sculptape.json; every PLACEHOLDER flag on the Analytics + Proforma pages clears on the next export; askst-proforma-costswithdrawn. Also: Today / Yesterday ranges with an hourly series (Meta hourly pull) and percent / $ input fields on the Forecast and Costs tabs. - 2026-09-19 — SUBSCRIPTION ANALYTICS LIVE on the workspace (Analytics tab) — David judged the wireframe (Lane A metric grid · all five splits · console-ask alerts · slider forecast · a Costs tab with a form + PDF drop) and it was built the same day:
tools/analytics/(SQLite warehouse from 29next orders/subscriptions/transactions/users + Meta ad-level insights; PII-free facts; the app computes MRR, lifecycle, churn by lane/cycle, subscription + customer cohorts, the TBYB funnel, LTV/CAC/payback by campaign · ad set · ad · prelander · lane · zone · source, P&L + MER, products/lanes/basket, RFM segments + CSV, the forecast, alerts)./api/costsstores the Costs-tab numbers + dropped PDFs (→tools/analytics/inbox.md). First alert fired on sync: 3-day CAC above $90 (console ask). The checkout-URL params now reaching 29next (utm_campaign = campaign id, utm_content = ad id, utm_term = ad set id, pl/plid) are the join that makes LTV-by-ad possible. - 2026-09-19 — PROFORMA (playbook Phase 8) LIVE on the workspace (David: "the next phase in the playbook is to do a proforma based on the data that is coming in from facebook ads and conversion data and revenue…").
tools/proforma/— Meta campaign spend (act_243417265464399, campaigns matching 'Sculptape') + 29next orders/subscriptions (skinscience store, ST- prefix, test orders out) → lanes from the package names → every order priced againsttools/proforma/assumptions/sculptape.json→/sculptape/proforma/on the Launch Center: daily P&L, lane mix, attach, cohorts, observed-vs-assumed, 12-month forward model with editable inputs. First read (4 days): spend $641, 4 paying customers, CAC ~$160, AOV $29, first orders 2× ladder-2 / 1× ladder-1 / 1× TBYB-sub, bump attached on 2, VIP on 1. COGS, freight and the Lansil rate card are NOT on file — every cost on the page is a PLACEHOLDER ($3.50/pack COGS, $0.40 sea freight, $2.50 pick/pack, $0.60 packaging, $4.75 postage) → console askst-proforma-costs. - 2026-09-18 — URL-parameter forwarding gap on the ?ob=c lane FIXED (David: "the url parameter forwarding isnt working from prelander to pdp"). Traced live: prelander → PDP was intact (the forwarder stamped utm_*, fbclid, fbp_in/fbc_in/eid_in, plid/pl/cta on every CTA at click); the drop was PDP → checkout on the Try Before You Buy CTA + sticky bar — the OB-C script built bare
/checkout/?pkg=N&ob=c&plan=kandsyncCheckoutLinksskips#obcCta/[data-obc]on purpose, so since every prelander started landing on?ob=c(2026-09-17) the paid funnel lost all params at that hop and 29next recorded no UTM on those orders. Fix b63578073:obcHref(k)= pkg/ob/plan first + the PDP's own params, static CTA href stamped at init; six PDPs +build_pdps.py. Verified live (webkit iPhone): checkout URL via the TBYB CTA now carries utm_source/campaign/content, fbclid, fbp_in/fbc_in/eid_in, plid/pl/cta; drawer lane unchanged; tryfirst audit 12/12, quick matrix 74/74. Note: the 60 active ads carry NO url_tags (58/60 empty) — only Meta's fbclid reaches the prelander, so per-ad UTMs in 29next need url_tags on the ads (adscaler). - 2026-09-18 — FULL ATTRIBUTION AUDIT (David: "do a full attribution audit for sculptape, from the playbook, since we have live traffic finally") — 5-phase protocol run; 3 real gaps found + fixed + re-verified 15/15. Phase 1 (code): namespaces clean (pages bind ST_+NS_, runtime aliases both), no manual _fbc/_fbp writes, Pattern I + M present, SDK facebook provider off, external_id pre-hashed — but (a)
st-cart.jsAddToCart was a rawfbq()with no event_id / no CAPI twin, (b) the runtime fired PageView only on PDPs. Phase 2 (live): runtime + forwarder + pixel 1065595576190917 on PDP and prelander, /api/capi 400-probe, unsigned webhook 401. Phase 3 (tools/qa/sculptape_attribution_e2e.py, fresh Chromium, prelander → CTA → PDP → drawer → checkout, every /api/capi body + pixel hit captured): baseline 8/15 — ZERO ViewContent anywhere (Meta 24h volume confirmed: PageView 1851, IC 135, ATC 63, ViewContent 0), drawer AddToCart browser-only without event_id; identity chain itself was intact (fbp/fbc/eid forwarded at click, same external_id hash on all three pages, PageView/InitiateCheckout browser+server same event_id). Phase 3.5 (29next): webhook 166 Active, warning_count 0, order.created only, target repointed from sculptape.netlify.app to https://sculptape.com/api/order-webhook; store also carries SS's 29OrdersCapi (100) + PS relay (199) + 'Nexus Report Webhook' (1, all events, external) — the ST- SKU gate keeps them apart. The one REAL order since launch (101717, 2026-09-16, ST-06, $0.00, canceled, funnel null, empty SDK metadata) did not come through the checkout — admin/manual. Phase 4 (Graph API): pixel healthy (first-party cookies on, AAM on with external_id), 24h split SERVER 73 % / BROWSER 27 % — dominated by QA runs (headless browsers are bot-blocked by fbevents, CAPI still lands) and the day's 3 pixel Purchases were the 3 QA sandbox orders (the relay had no is_test gate); ad account act_243417265464399 has spent $2.53 / 66 impressions / 1 click on 'Sculptape (ST-01) — All' — too thin for EMQ, re-read Events Manager Diagnostics after 48 h of spend. Phase 5: GA4 read needs ADC (not set up here) — manual Realtime check. FIXES (f581388a9, deployed): PDPview_item/ViewContent in st-attribution.js (browser + CAPI, one event_id, content_ids = zone SKU); drawer AddToCart throughST_TRACKING.trackEvent;shared/functions/order-webhook.mjsignoresis_testorders unless FB_TEST_EVENT_CODE routes them to Test Events (signed probe → 200 'Ignored: test order'; foreign SKU → 'Ignored: foreign-brand order'; unsigned → 401). Re-run 15/15 (resultstools/qa/results/st-attr-20260918-0828). Other brands pick up the is_test gate on their next deploy. - 2026-09-17 — FULL MATRIX QA (David: "lets run full matrix qa on sculpape for the tbyb lane, and full price lanes, on all platforms") — 414/414 rows green: chromium Android + desktop, webkit iPhone + iPad + Safari desktop = 345/345 first pass (st-matrix-20260917-2204); firefox desktop 69/69 on the re-run (st-matrix-20260917-2219) — its 15 first-pass reds were a Playwright artifact (navigator.language = 'undefined' without a context locale → the 29next SDK's Intl.NumberFormat throws RangeError; real Firefox has a locale), plus fbevents.js's cookie domain-probe warning that Firefox logs before the cookie lands (verified present).
tools/qa/sculptape_cart_matrix.py(port of the CellScience matrix): chromium Android 412 + desktop 1440 · webkit iPhone 390 + iPad 820 + Safari desktop 1440 · firefox desktop 1440; rows = drawer (D1–D5: picker + Add to cart intercept, bag count, +/−/Remove, $60 gift threshold, seed → /checkout/?stcart=1 with utm+fbclid), checkout modes (C1 cart mode collapsed summary = SDK total; C2 ?pkg=3 purge; C2z ?pkg=9 zone remap), OB-C on all six zones (O1–O4: column swap, sub/one plan picker, day-7 foot copy, session-sticky, drawer bypass), C3 all six zones × both TBYB lanes (TB card on shipping 2, 'Try First — <Zone> (1 Pack)', $0.00 line + keep price EVERY 7 DAYS, shipping $6.99, total $16.98 with the bump on), C4 empty checkout → /collection/, X1 PayPal express handoff, T timer rate, Z console. Resultstools/qa/results/st-matrix-20260917-2204 + -2219. **Dummy order (tools/qa/sculptape_tbyb_order.py, webkit iPhone, sandbox Discover): order 103072 is_test ✓ — lane line $0.00, Shipping Protection $9.99, shipping $6.99, VIP (pkg 51, $12.49) accepted, upgrade + bundle skipped, thank-you clean; back office: day/7 subscription #298 at $19.99 with NO shipping riding (relay 8c works), VIP day/30 #299 at $9.99; both cancelled by the runner. Webhook replay skipped (no ~/.sculptape.env backup — the relay was proven 2026-09-14). Notes for David: (1) the shipping-protection bump rides ON on the trial lanes ($16.98 today, Charm pattern; CellScience runs it unselected) — a pricing call; (2) the country select shows 10 countries, default US. - 2026-09-17 — TBYB keep charge moved to DAY 7 (was 14) — David: "its pay on Day 7 not 14 for sculptape". 29next pkgs 5/6/11/12/17/18/23/24/29/30/35/36
interval_count14 → 7 viabuild_29next.py --align(the align routine now comparesinterval_count, it only comparedintervalbefore); copy flipped in build_pdps/build_checkout/pages_content/check_compliance and the built PDPs, checkout, Terms §3.2, Refunds. Redeployed. - 2026-09-17 — Every PDP image broke twice after the checkout hotfix. (1) The hotfix deployed sculptape.com from a worktree:
shared/cms/(266 gitignored slot images) is absent there → shipped with zero CMS images; redeployed from the main checkout. (2) Even then 84 sized variants (-sm.webp/-md.webp, PSI Layer 1.5 of 2026-09-07) 404'd — they had only ever existed in Netlify deploys, never on the main checkout's disk. Restored by listing deploy6aa8d6da…'s files and downloading the missing 84 from its permalink intoshared/cms/(350 files on disk now), redeployed; full sweep = 0 failed images on home, collection, six PDPs, checkout, OTOs, thank-you.tools/deploy-sculptape.shnow refuses a worktree or an empty shared/cms. - 2026-09-17 — Checkout summary default = $0.00 (David: "make the default not be $68 but $0.00 for all brands"). The order summary's baked placeholder (Grand Total
$68.29, Discount- $59.90— the checkout template's demo figures, shown until the 29next SDK fillsdata-next-display) is now$0.00on every checkout built from template/checkout/02 (Sculptape, HairScience, Memo, NailScience + the template itself; Charm/CellScience/PeptideScience already baked $0.00). Redeployed. - 2026-09-16 — Rounds 1+2 re-rendered NATIVE (41 statics, no branding) after David's 'all beige' verdict; law recorded in §6, memory, SKILL §VIII.
- 2026-09-16 — Round 2: 28 angle-scoped statics generated from the swipe keeps × framework library; on the J&T board (44 with r1 + videos).
- 2026-09-16 — Hub statics board retired; ads judged on the J&T board only. Static-image Ads Library sweep r1–r3 for body-tape started (angle-aware swipe library).
- 2026-09-16 — J&T ad pipeline wired (board + angle-aware swipe library, registry entries, r1 statics + Studio videos on the board); ads playbook §VIII documented.
- 2026-09-16 — Top-5 thigh prelander statics r1: 13 angle-scoped statics boarded (
sculptape--top5-statics-r1), ask st-top5-statics-r1. - 2026-09-15 — Angle Studio round 1: 13 angles + 65 hooks boarded (
sculptape--angles), body-tape ad sweep + Deep Research VoC banked, console ask st-angle-board-r1. - 2026-09-14 — Guarantee → 30-Day Money-Back Guarantee across site, six PDPs, checkout, legal, thumbs and the six top-5 prelanders (was 60-day Love-the-Mirror); registry
guarantee_days60 → 30. - 2026-09-14 — TBYB renewals ship FREE + relay env RESTORED (from the CellScience lane): the production site had only FB_*/GA4_* env — no NEXT_WEBHOOK_SECRET or ORDER_SKU_PREFIX, so
/api/order-webhookran UNSIGNED and un-gated (an unsigned POST returned 200; every skinscience-store order — SkinScience + PeptideScience too — was eligible to forward as a Sculptape Purchase). Set from webhook #166'ssecret_key+ST-, plus NEXT_ADMIN_BASE/KEY (skinscience store) and TBYB_SHIP_FREE_LANES (12 zone lanes; step 8b fulfilment automation OFF). Shared relay step 8c + hourlytbyb-shipping-sweep.mjsdeployed viabash tools/deploy-sculptape.sh. No live trial-born subscriptions carried shipping. Root cause of the missing env: unknown — the 08-2x accounts-API set did not persist; verify env withnetlify api getEnvVarsat every launch gate. - 2026-09-02 — created from template + imported 2 memory file(s) as the starting record.
- 2026-09-06 — §6: editorial voice is a beauty editor (prelanders re-worded).
- 2026-09-05 — §5 comparability rule + §6 approved rival rosters (six Top-5 prelanders rebuilt on the comparable-only roster;
prelanders/sculptape/build_top5.py).
- 2026-09-21 — AddToCart on the Try Before You Buy CTA (David: "yes fire add to cart on the tbyb button across brands"). The trial CTA (
#obcCta+ the sticky bar in ?ob=c) links straight to /checkout/, so the lane never produced an AddToCart — Ads Manager read ViewContent → InitiateCheckout with an empty Adds-to-cart column (seen on Sculptape's RS – Spare campaigns).tracking/st-cart.jsnow fires AddToCart through the runtime on any trial-link click (plan=tbyb-/ob=ccheckout hrefs; event_id + CAPI twin via sendBeacon; value = the plan's day-7 keep price $19.99 / $24.99; 1.5 s throttle per link) and the link proceeds untouched. Drawer AddToCart unchanged; no PDP rebuild — /tracking/ is cached ≤ 10 min. Skill: attribution-stack Pattern N; spot-checktools/qa/tbyb_atc_spotcheck.py. - 2026-09-10 — Prelanders MOVED to EcomScalers (David):
sculptape-prelanders(616390cc-ab0d-4973-81d2-5887565914de) was still sitting on the personal "David Sanghera" free team while sculptape.com had been on EcomScalers since 2026-08-27 — the six Mirror Test pages were the last Sculptape surface off the house team. Moved viaPOST /api/v1/sites/<id>/transfer{"account_id": "696867a8453db489e0351131"}(the only route that works —PATCHwithaccount_slugreturns 200 and silently ignores the field). Site id, name andsculptape-prelanders.netlify.appare unchanged, so every manifest live_url, ad link and Copy Studio reference still resolves; no custom domain or DNS zone was attached, nothing to re-point. All four site env vars (FB_PIXEL_ID, FB_CAPI_ACCESS_TOKEN, GA4_MEASUREMENT_ID, GA4_API_SECRET) travelled with the site byte-identical, verified against a pre-move backup. Plan wentnf_team_dev→nf_team_pro(deploy retention 90d). Redeployed throughtools/deploy-st-prelanders.shwith--skip-functions-cacheso the three functions (capi, order-webhook, slot-proxy) rebundled under the new team: CDN diff requested 0 files, i.e. content byte-identical to what was already live. Verified after: six pages 200,/api/capirouting,/slot-proxy?page=200, and the stored pixel+token accepted by Meta (events_received: 1, TEST5949). Served HTML shrank 536 bytes/page — that is Netlify's own injected HUD script switching to its smallervariant=publicform on Pro, not lost content. STILL OFF THE HOUSE TEAM:sculptape-launch-boardon Riptide — David's call whether the external board moves too. - 2026-09-20 — NEW META AD ACCOUNT:
act_268669251366940"RS – Spare" (BM AMV LLC113824746806442) — David: "new ad account is - 268669251366940 for sculptape… the pixel is connected to it". Verified from the pixel side: pixel1065595576190917(owned by DeltaSift / Liaison–B) is shared to AMV LLC and assigned to RS – Spare, so ads there optimise on it (anact_X/adspixelslisting does NOT show shared pixels — read<pixel>/shared_accounts?business=instead); the Carol Anne Foster page927552797118674is owned by AMV LLC. Liaison–B's Sculptape campaigns (All · Video / All · PDP ob=c / Question posts) are PAUSED after $294 on 2026-09-20; RS – Spare carries no Sculptape campaign yet and hosts other advertisers' Spanish-market campaigns (OptiClear, AhorraLuz, CeroRayas, BrisaFría — ~$870/day), so EVERY pull from it must filter by campaign name. Repo:tools/proforma/assumptions/sculptape.jsonlists BOTH accounts undermeta.accounts(Liaison–B = the Sept 17–20 launch history, RS – Spare = go-forward);proforma.py/analytics.pyiterate the list withcampaign_matchper account;tracking_head.pycomment +jt-launch-post.pyexample updated. adscaler picks the account per campaign (Campaign.fbAdAccountId) from the system user's accessible accounts — the next launch selects RS – Spare, nothing to migrate. Scale OS deliberately has NO config for RS – Spare: its configs are per-account with no campaign filter, and on a shared account it would act on other advertisers' campaigns. WHY (David 2026-09-20): "we switched the accounts since the old account had expensive CPMs" — Liaison–B's CPMs were the problem, RS – Spare is the cheaper-CPM account. Watch after the first RS – Spare launch: CPM by account on the Analytics tab (meta_ad_daily rows carry the campaign, so the account is read off the campaign's home) — if RS – Spare's CPM is not materially lower on the same creative, the move did not buy anything. OPEN for David (BM clicks, not code): confirm shapeandsculptreviews.com + sculptape.com are verified domains in AMV LLC (our token cannot read owned_domains);act_268669251366940/promote_pageslisted 0 pages for our system user — check the page is assigned to the ad account before the first launch.
2026-09-27 — STORE FIX (money): the twelve BOGO packages 64–75 were priced $24.99, qty 1, since their creation on 2026-09-22 — the create-ignores-prices trap (prices on POST is silently dropped; the package inherits the $24.99 variant price) and --align never ran after that create. The SDK therefore charged $24.99 (+ $9.99 protection) for "Buy 2 + Get 1 FREE — 3 Packs" AND for "Buy 3 + Get 2 FREE — 5 Packs" while the Total Order Summary rows said $49.98 / $74.97 (its total read the SDK's $34.98). Reachable on the live checkout only from the 2026-09-27 deploy that carried the BOGO ladder; exposure = 0 orders (5 orders on the skinscience store since 2026-09-22, none with a BOGO line — admin read 2026-09-27). Fixed the same hour: python3 products/sculptape/build_29next.py --align → 12 × PATCH 200 ($49.98 / $74.97, is_recurring:false sent explicitly), read back through the client Campaigns API: all 12 correct. Rules from it: (1) every package create is followed by --align + a client-API read-back in the same turn; (2) a checkout gate compares the SDK total (next-cart-state totals) with the summary rows — a mismatch is a store bug, never a display nit.
2026-09-27 — checkout zone remap bug (live since the same deploy): /checkout/?pkg=66 and every non-silhouette direct link kept the silhouette BOGO cards 64/65 and re-pointed only the 1-pack — the pre-SDK remap (silIds) still named the retired 1/2/3 stamped cards. Fixed in build_checkout.py: the map is generated from the stamped silhouette ids ({1:0, 64:1, 65:2} → the zone's [1 Pack, BOGO 3, BOGO 5]). Pre-deploy proof: local checkout intercepted over sculptape.com, seven zone links → cards re-pointed, zone thumbnail, cart = the selected id.
2026-09-27 — radio-row selector + summary under the pay button + guarantee badge + no sticky price LIVE (tools/deploy-sculptape.sh from /dev): PDPs from products/sculptape/offer_box.json (layout radio; build_pdps.py stamps the zone's TBYB id into the trial row and runs pdp_house.guarantee + hide_sticky_price); checkout via tools/checkout_selector.py (runtime map of all 18 zone ladder ids + thumbnails, the stamped-id zone remap, the trial card without its empty image column); deploy script no longer ships offer_box.json. Pre-deploy proof: brand check 60/60 (WebKit phone + Chromium desktop: seven zone links re-pointed with the zone thumbnail, TBYB sub + one-time lanes, six PDP trial rows → checkout), offer_box_check 6 PDPs × 2 viewports 0 failures, deploy hygiene PASS. Seed race found on the way (house-wide): the SDK's PackageSelectorEnhancer seeds its pre-selected card only into an EMPTY cart; with the bound display elements gone from the restyled cards the pre-checked Shipping Protection toggle initializes right behind the selector and won in ~half of WebKit loads (cart = protection only). Fixed in the shared module (sel3-seed: on a ?pkg= link, once the SDK is up and the cart holds anything but the linked package, add it once + re-apply the card's shipping method) — seed rate 6/6 on the TBYB and zone links after.
Imported status record (2026-09-02) — distribute these facts into the sections above as the brand is touched
From memory sculptape_cart_drawer_status.md
Rule (David 2026-08-26): any brand with SKU assortment > 1 gets a slide-out cart. Sculptape = first live implementation. Chassis to clone = products/nailscience/cart-drawer-preview/index.html (mock lane) → products/sculptape/tracking/st-cart.js (functional lane). Never hand-build a drawer.
LIVE 2026-08-26 (approved, gift ON):
st-cart.json all 6 PDPs: intercepts tier checkout links (capture phase; TBYB#obcCta/plan=tbybpasses through), floating chip + nav#cartOpentakeover, retires the template's dormant single-item mini-cart (#cartDrawerremoved at boot), cross-zone rail, gift = Free Bonus Pack of chosen zone (pkgs 57–62, real $0 packages carrying zone variants, ships from warehouse) unlocked at $60, bump 37 seeded pre-checked. localStoragest_cart_v1.- THE ONLY checkout
/checkout/(David consolidated fully) — post-processed from the built silhouette checkout instamp_brand_checkout(); THREE modes: cart (?stcart=1: selector hidden, lines panel),?pkg=selector (three cards remapped to ANY zone's ids pre-SDK-boot + selected; heading flips back to "Choose Your Package"), TBYB all zones (TB map extended viaObject.assigninjection at theif (p && TB[p])anchor). Guard allowlist = all 61 ids; empty direct visits bounce to/collection/(skipped when?pkg). Zone checkouts + non-sil OTO chains/thank-yous are BUILD INTERMEDIATES ONLY — deleted in main(), 301'd in netlify.toml (/products/:slug/checkout/*→/checkout/; zone special-offer → sil chain; zone thank-you → sil thank-you; Netlify file-first means sil's real pages shadow the :slug rules). ONE OTO chain (silhouette path) serves every purchase — dynamic per-zone routing REMOVED; cart-content-conditional upsell routing = the open problem David flagged. - Handoff preserves
location.search(utm/fbclid/…) — dropping it re-opens the "No attribution found" gap.
campaign-cart SDK contract (v0.3.10, empirically verified live):
- Persisted cart = sessionStorage
next-cart-state, zustand shape{state:{items:[...],appliedCoupons:[],subtotal,…},version:0}; items carry packageId/quantity/price/title/is_upsell/sku/productId/variantId…. - Seeding that shape pre-load = the multi-item API: SDK rehydrates, enriches, recalculates totals, accepts $0 lines.
window.nextReady.push(cb)= ready hook; publicnext.addItem({packageId,quantity})exists. - LAW: restored state ALWAYS beats
?pkg=and selector defaults. Pre-existing bug: bounce-back shoppers who re-picked a bundle were charged the stale cart. Fix (all checkouts):?pkg=withoutstcart=1purgesnext-cart-statebefore the SDK boots.
Express row (added on David's ask): drawer shows PayPal + G Pay always, Apple Pay only when window.ApplePaySession (mirrors SDK device-gating; empirically PayPal+GPay render on the campaign, Apple absent in Chromium). Buttons are real handoffs: seed cart → /checkout/?stcart=1&express=<m> → checkout polls for the SDK express container (setTimeout poll, no MO), scrolls it center + champagne pulse, re-asserts once at +1.2s (late SDK hydration resets scroll; also wait for window load +350ms).
Nav (David): logo center; cart bag top-right ALL viewports (#cartOpen, badge = data-stcount via CSS attr(), hidden at 0); Shop lives in the mobile sandwich menu (.hamburger toggles #navMenu, closes on outside click); floating chip REMOVED.
Gates/laws honored: compliance gate rewritten for single funnel (§3.4b brand checkout + express seek + st-cart express/ApplePay/no-new MutationObserver; §3.5 asserts zone pages REMOVED + sil chain intact; §3.55 TBYB on brand checkout + PDP /checkout/?pkg hrefs; netlify.toml 301 assertions). Gate greps for banned words must target new MutationObserver, not the word in comments.
SIM QA EXECUTED (2026-08-26, iPhone 17 sim, dummy Discover card — gateway is TEST MODE, orders 102179/102180 booked): full pass green — drawer/rail/gift/express (Apple Pay renders on iOS ✓), Places autofill, Spreedly iframes accept sim keystrokes, purge law, OTO chain, receipt. Caught + fixed live: (1) 29next TRAP — package CREATE silently rewrites price 0.00→variant price on NON-recurring pkgs (recurring 0.00 survives): first QA order charged the "free" gift $24.99; PATCH sticks (gifts 57–62 re-zeroed; order 102180 verified gift $0.00, total $134.97); always --align after create; gate asserts 0.00 via public campaign API. (2) css-combine loop shadowed name → receipt greeted "custom.css"; brand-phrase greeting now. (3) Checkout template shipped 90-day guarantee ×4 + two 90d.webp seals (text gate only swept PDPs) → all 60-Day Love-the-Mirror, seals removed, gate sweeps checkout. (4) 'shiping' typo. Webhook 166 active, warning_count 0 across both orders.
VIP-accept defect RESOLVED (2026-08-26): upsell pages loaded SDK config via ../../checkout/sculptape/config.js — the consolidation deleted zone checkouts, the 301 served /checkout/ HTML as 'JS' → SDK keyless → accept navigated without POSTing (decline-path QA can't catch it). Fixed to /checkout/sculptape/config.js; gate asserts no ../../checkout/ refs on OTO pages. Verified: order 102183 carries the $12.49 VIP upsell line + ACTIVE sub. TRAP: page-tree deletions must sweep surviving pages for relative refs INTO the deleted tree — redirects mask the break. Diagnostic pattern that cracked it: Spreedly-tokenize the dummy card via the public payment_env_key → create order via campaigns API → replay the upsell POST server-side (API 201 = client-side bug) → console on the page with a live ref_id. Browser-pane CANNOT click into Spreedly iframes (OOPIF) — sim or server-side only. Also: receipt page has no order-lines summary (minimal by design — confirmation email carries details). SECRET50 typeable voucher doesn't exist (no voucher API; chip = story pattern).
Traps: (1) own-CSS reset .stc button{...} (0,1,1) out-specifies .stc-go (0,1,0) — prefix component selectors (.stc .stc-go). (2) build_pdps tail-eater: the [23]-marquee rebuild slices t[vs_start:t.index('</body>')] — anything injected before </body> earlier in the pipeline is silently swallowed; inject tail scripts LAST (nav-menu toggle learned this). (3) In-memory drawer cart hydrates at page load — clearing localStorage after load doesn't clear the running cart (test-sequence artifact, not a bug).
CRO/readability pass (2026-08-26): WCAG-audited the real pairs — PDP --ink-3 was 2.44:1 (one token → every eyebrow/caption; now #726A5C 4.99); drawer mute→#756D5E, trust→#7E6136, pill→#6E5426/11px, gift tag→terra bg/10px (strikes stay faint by design). Gift <select> 12.5→16px (iOS zoom), steppers 42pt, close 44pt, empty state links /collection/. Checkout form audited CLEAN (16px/48pt fields, full autocomplete; tel-full→tel). Chrome-on-iOS class (David saw a difference): drawer 92vh→+92dvh (collapsing toolbars) + Apple Pay absent there = expected (no ApplePaySession — 2 express buttons, not 3). QA plan section L added.
Post-QA live fixes (2026-08-26 pm): (1) CMS slots 24–28 (results carousel) were uneditable — .ba-slide img{pointer-events:none} (drag smoothness) made slot imgs click-transparent; st-slots edit mode now re-arms pointer-events + capture-stops mousedown/touchstart (generic for all carousels); verified with a trusted production click. (2) Copy rule: "Zone"→"Area" in ALL customer copy (Multi-Area Body Tape, area-cut panels, six areas…) — code identifiers + store package names untouched; the built-output visible-text sweep (strip tags/scripts → regex) must report ZERO before deploy.
PDP IMAGE ROUNDS (David 2026-08-26): Round 1 = "easy" non-buying-decision images — slots 13/14/15 clinician avatars, 16/17/18 bundle cards (1/2/3 packs), 36/38/39/41/42/43 reviewer avatars (Sharon R./Megan L./Maria G./Angela H./Rosemary C./Cris M. — 37/40 are review PHOTOS, separate). Round 2 = Hero Gallery (slots 01-06 + thumbs 07-12), ONLY after Round 1. Round-1 candidates GENERATED (22: 6 clinicians, 4 bundles i2i-anchored on the locked box render, 12 avatars — fresh unique faces per reviewer-avatars doctrine, NEVER the shared av-*.jpg template faces) → verdict board sculptape--round1-easy-images live on the launch center (assets served same-origin from /docs/round1/). AWAITING DAVID'S PICKS → winners wire into the slot CMS via /api/slots. AI-people note: David explicitly green-lit AI generation for these avatar/doctor slots (his no-AI-people rule stays for hero/worn imagery — photography lane).
ROUND 1 WIRED (2026-08-27): David's picks → slots on ALL six PDPs: clinicians 13/14/15 = clin-2 (man 40s) / clin-3 (SA woman 30s) / clin-4 (EA man 50s, glasses); avatars 36 Sharon-car · 38 Megan-mirror · 39 Maria-café · 41 Angela-beach · 43 Cris-gym; 42 Rosemary was UNPICKED → defaulted to rosemary-a-garden (one-click swap if he wants dinner). STANDING RULES from David: (1) clinicians + UGC avatars REUSE across every PDP; only the reviewer NAMES change per product — builder ZONE_REVIEWERS map swaps the canonical six (+Cris D.) per zone (Waistline Dana W./Kelsey T./Elena R./Tasha B./Joanne F./Mia S.; Lift Carla J.…; Neckline Susan K.…; Profile Diane M.…; Sleeve Wanda S.…). (2) Bundle cards = BOXES ONLY, no tape rolls/props — the approved 1/2/3 compositions (upright single; two side-by-side one forward; three in one straight row, front-facing, evenly spaced) rendered per zone via i2i on each zone's locked box render (identity/images/w3-st0X). Generated with Gemini 2.5 Flash Image called DIRECTLY (scratch gen_bundles.py; token in repo .nano-banana-config.json) when the nano-banana MCP dropped — same model, imageConfig.aspectRatio + responseModalities:["IMAGE"]; QC every batch on a contact sheet (traps: 4th box sneaking in, garbled side-panel text, diagonal gold stripe artifacts, tape slivers — regenerate, ~20% redo rate). Files: creative/sculptape/gallery/round1/bundles-boxesonly/. Round 2 = Hero Gallery next.
SITE CHROME STANDARD (David 2026-08-26): one nav + one footer on EVERY page, PDP style is the source. products/sculptape/site_chrome.py = single source: header_html(links, announcement) (PDP-identical markup: announcement w/ SECRET50 timer, hamburger/logo/bag, nav-links, #navMenu sandwich, #cartOpen), footer_html() (dark 4-col: all six products + The Line / Support / Company / Legal / pay icons / cosmetic-accessory copy), self-contained CHROME_CSS (literal PDP colors, sticky desktop nav ≥960px), CHROME_JS (timer + menu + loads st-cart.js), seo_head() (canonical+OG+twitter), breadcrumb_ld(). Home + collection + 7 legal pages rebuilt on it (semantic header/nav/main/footer; legal pages omit the promo banner); PDPs emit their footer FROM site_chrome (template [33] block replaced by index-slicing to </main>; parity CSS for link colors/pay-icon rule). Non-PDP nav links = site-level (Shop · The Silhouette · How It Works → PDP#how · Guarantee · FAQ → PDP#faq); PDP nav keeps in-page anchors. st-cart.js now on every page (bag/drawer sitewide). Checkout/OTO/thank-you deliberately keep their minimal funnel headers (CRO) — footer parity there is a follow-up if David wants it. New brands: copy site_chrome.py + the builder wiring.
Drawer top standard (2026-08-29, v1.7.8): header = Your Cart · check Free Shipping until <today+5d> · x (dated badge computed at render; no item count, no trust strip). ONE unlock module: gift progress bar, single 🎁 end-milestone, fill = sub/threshold. Footer: no Bundle-savings row; guarantee = "60-Day Money Back Guarantee. No BS." Icon-led copy rows = display:block + inline sized SVG, never flex (flex splits prose into columns). st-cart.js at ?v=9 — bump ?v= every edit (600s edge cache).
From memory sculptape_launch_status.md
Sculptape — women's body-sculpting tape, NEW BRAND, launched via the
Launch Console (🧭 :8788) orchestration. Domain sculptape.com (bought
2026-08-24). Name locked AGAINST sweep advice — Tarte owns "SCULPT TAPE"
(live, IC 003) but their recitation is "cosmetics **not in the nature of
tape**"; usage law in briefs/sculptape/name-sweep.md: one word
SCULPTAPE always, zero Tarte adjacency, NO PURPLE ever, body-wear not
color-cosmetics.
Truth Gate: 22 TRUE / 18 BANNED claims — briefs/sculptape/truth-gate.md
is the ONLY claims source (science layer = banned; register = "makeup for
your body"). Copy cites claim IDs.
Design system ALL LOCKED (board :8769/_product-jt/#brand=sculptape):
D5 Champagne Luxe (ivory #F2EDE3 + champagne foil #C9A96A + charcoal;
linen-greige photography) · tape is CLEAR (David's law) · FAM-B
pure-ivory glyph family · 6 SKUs ST-01..06 (Legs+Booty=HERO, Tummy, Boob,
Neck, Face, Arm; Boob/Face compact cartons). Spec:
creative/sculptape/LOCKED_SPEC.md + VISUAL_DNA.md (trap-baked prompt
template).
Geometry = SOURCED FROM TONEE (rips in creative/sculptape/reference/tonee/):
pre-cut clear film PANELS on backing, zone-specific — NEVER rolls (roll
prop retired). FACE = string-lift kit (petal anchors + cords → different
BOM). Tonee intel: entire store Out Of Stock (2026-08-24); their glyph
pouches validate FAM-B instinct — ours must stay visually distinct.
**HARD lane split (David 2026-08-24): humans = photography team, AI =
packaging/product/flat-lay/macro only.** Shoot brief:
briefs/sculptape/photo-shoot-brief.md; wave-4 worn renders = comps.
Offer LOCKED: $29.99×1 · $24.99/ea×3 · $19.99/ea×5 uniform across
zones + SUBSCRIPTION AT LAUNCH — checkout defaults = OFFER BOX A + C (NOT B; David 2026-08-24)
(briefs/sculptape/offer-economics.md). Phase-5 traps queued: 29next sub
pkgs must exist before OB-B; Terms §3.1 auto-renewal. Open: Sub&Save %.
LOCKED late-adds: ST-03 Boob = NUDE (the one clear-law exception; modesty=function). ST codes INTERNAL ONLY; customer names LOCKED (System A editorial): ST-01 The Silhouette (HERO) / ST-02 The Waistline / ST-03 The Lift (nude) / ST-04 The Neckline / ST-05 The Profile / ST-06 The Sleeve — names on all customer surfaces incl. 29next display names. OPEN FLAGS:
whats-inside flat-lay render needs regen (window-frame drift) · Sub&Save %.
Launch-console habits: check python3 tools/launch-console/console.py answers
every launch turn; verdict marks auto-save ([[launch_console]]); board
verdicts at creative/_product-jt/verdicts/sculptape.json. Sitemap =
Shopify-index gold standard, generates from
creative/sculptape/pages-manifest.json at deploy, launch gates on
tools/automation/build-sitemap.py --validate ([[product_launch_skill]]).
Next: PDP copy from TRUE claims → PDP on Base Template 1 → checkout
(18+ pkgs) → tracking → compliance → launch.
2026-08-24 PDPs LIVE ON STAGING: all six built via
products/sculptape/build_pdps.py (config-driven stamp from Base Template
1; re-run after template/copy changes) → sculptape.netlify.app (site id
c9d5d07c-037b-43e5-99fa-1a70dcedbe05, root 302→the-silhouette). Shared
assets at /shared/ (deduped after the bulk-push wedge). Verified live: 0
clone leaks, GTM/pixel stripped (Phase 6 rewires), ?pkg handoff live with
PLACEHOLDER package ids. Placeholders flagged: AI comps for photo team,
review photos, 60-day guarantee (unconfirmed). GATES: David's 29next store
step (campaign + 6×{1,3,5} pkgs + subs) → checkout build; DNS cutover →
sitemap validate gate.
EXTERNAL LAUNCH BOARD (standing): https://sculptape-launch-board.netlify.app
(Netlify site 928f3649-9a05-484e-aaa9-8e3d05cb56cb) — David's rule: launch
visibility for people WITHOUT console access. Generator
tools/launch-board/build.py --brand sculptape + curated external-safe
config tools/launch-board/sculptape.json (PDP links, ownered next-steps,
photo brief; NEVER legal analyses/internal economics). **Redeploy on every
milestone** — update the config's next_steps/done as phases move.
2026-08-24 (late) FULL SITE + OB SYSTEM: homepage live at root (real
page; redirect retired) · collection + 7 legal pages all carry nav +
canonicals · JSON-LD everywhere (Product+Breadcrumb per PDP, ItemList
collection, Org/WebSite home) · PDPs = numbered CMS-slot skeletons (46
slots/page + V1-V7 marquee "Real Results. Real Bodies.") · **OB-A + OB-C
LIVE** (?ob=c TBYB: sub $24.99-day-14 / one $29.99-day-14; OB-B excluded;
sub keep-price = placeholder pending Sub&Save %) · value badges ON bundle
images · SHOP nav button · copy spec measured (2-line subhead, 2x4-line
problem+solution). Logo wave 6 on board (8 SVG concepts) AWAITING PICK.
**The consolidated future-launch doc:
skills/product-launch/phases/new-brand-track.md** + vault note
"Product Launch — New-Brand Track + Launch Console". Builders re-stamp
everything: build_pdps.py + build_site_pages.py (edit config, rerun,
redeploy).
2026-08-25 29NEXT BUILT (fully automated): OAuth app "Claude
Sculptape" (id 133, scopes admin+campaigns+catalogue r/w) · token in
~/.29next/skinscience.env (Bearer; David will rotate) · **campaign
"Sculptape" id 1838** (client api_key in creative/sculptape/29next-map.json)
· products 136–146 (ST-01..06, variants 137–147) · **36 packages
store-truth verified**: per zone 1/2/3-pack $24.99/$37.50/$44.99 +
Sub&Save $19.99/30d (PLACEHOLDER %) + TBYB $0→day-14 keep. Builder
products/sculptape/build_29next.py (report/--create/--align, idempotent).
API traps: prices write as prices[] (create IGNORES them — align via
PATCH); is_recurring DERIVED from interval (null=one-time); Bearer not
Token; package list = bare array; 429s need backoff. PDPs carry real ids
(QTY_TO_PKG per zone; OB-C hrefs carry tbyb pkg ids). NEXT: checkout chain
(template/checkout/02 port, campaign key 1838), then Sub&Save % re-align.
2026-08-25 CHECKOUT LIVE: six zone checkouts at
/products/<slug>/checkout/ (build_checkout.py; campaign 1838; bump 37;
CORS-proxy /next-api/* + fetch shim; ?pkg preselect; zone-cart guard) +
thank-you pages (tracking-free, Purchase-not-fired-here). VERIFIED:
hydration, preselect, cross-zone reseed, payment iframe. PDPs link
relative checkout/. REMAINING: Phase 6 tracking (pixel/CAPI/GTM/attribution
runtime + TEST5949 E2E), Sub&Save % re-align, TBYB checkout-lane handling
(?pkg=tbyb ids currently fall to default — OB-C plans need checkout-side
support), guarantee confirm, logo pick, DNS, photo shoot. OPTIONAL: add
sculptape domains in campaigns-app Allowed Domains for native wallet flows.
2026-08-25 PM — CHECKOUT RESTYLE + UPSELL CHAIN LIVE: checkout bg =
PDP paper #FAF7F0 (new hard rule file), honest card anchors ($49.98/$74.97
strikes, 25%/40% pills, LAUNCH40 chip replacing SECRET60, bump $19.99 fake
anchor removed) — strike/pill elements made STATIC (29next can't hold pkg
retail; SDK zeroes bound elements on the selected card). FULL OTO CHAIN:
checkout → special-offer/upgrade (per-zone Restock 3 Packs $29.99 one-time,
pkgs 38-43) → special-offer/vip (Sculptape VIP $9.99 + $9.99/30d, pkg 44,
product 148 ST-VIP requires_shipping=False, TEXT-ONLY page) → thank-you.
SDK v0.4.15 upsell contract on PS/HF chassis; Terms §3.1 broadened to
cover VIP renewal; compliance gate now 20 customer pages incl. upsell
wiring/disclosure/text-only assertions. **VIP promises David must honor
operationally:** member restock $19.99/pack + free priority shipping +
cancel-link in order emails. **NEW #1 GATE: campaign 1838 has ZERO
shipping methods** (SDK errors "Shipping method 1 not found"; not in the
store admin API — only in the campaigns-app UI): add method default
$0.00 like PS campaign 1748. TBYB ?pkg lanes on checkout still open.
**2026-08-25 EVE — OTO CHAIN = HF STANDARD (David: "OTO1/2/3 taken from
HumanFrame").** Chain now mirrors products/humanframe/shoulder/special-offer:
checkout → OTO1 vip (HF chassis) → OTO2 upgrade (HF oto-* page:
announcement/navbar/hero-block/offer+badge/reviews — css extracted from HF's
inline styles, champagne-swapped, lives at products/sculptape/oto.css) →
OTO3 bundle (HF cheap-add-on slot → cross-zone first pack $14.99/40% off,
pkgs 45-50, PAIR map: sil↔waist, lift→sil, neck↔profile, sleeve→sil) →
thank-you. Honesty deltas vs HF: no "double-bill" fear copy (gate asserts
its absence), no fake stock counters, reviews reuse the 3 house PDP reviews.
Compliance gate = 26 customer pages.
**2026-08-25 NIGHT — VIP = HF OFFER FORMULA (David's law, all future
launches): OTO1 VIP = 1× SKU unit + 1× VIP Customer Perks, offer price =
50% off the SKU ($24.99→$12.49), then $9.99/mo.** Per-zone VIP pkgs 51-56
(price 12.49 / recurring 9.99 / day-30) store-verified; membership-only
pkg 44 + product 148 RETIRED (left in store, unreferenced). Box mirrors
HF: value stack w/ right-spans, Retail vs Offer rows, "Complete My Order".
SHIPPING GATE CLOSED — David attached store method Default/$0.00 to
campaign 1838; SDK resolves {id:1, code:default, price:0}, cart totals
clean. Shipping automation findings: store registry READABLE at
admin/shipping-methods/ but POST=403 (scope missing — next launch check
the OAuth app Permission page for a shipping scope); per-campaign attach
lives in the campaigns-app UI only. Since shipping IS free now, the
stripped FREE-SHIPPING pills could honestly return — David's call.
FREE-SHIPPING pills restored (David 2026-08-25): checkout cards read
"Save 25%/40% + FREE SHIPPING" + plain "FREE SHIPPING" on the 1-pack —
static text; check_compliance now live-verifies a $0.00 campaign shipping
method whenever the claim appears (skips offline).
2026-08-25 LATE — PHASE 6 TRACKING STACK: GA4 HALF LIVE E2E. GA4
property 551617304 / stream G-DXG8TX4GW3 (created via Admin API —
acknowledgeUserDataCollection attest required before MP secrets; parent
acct = wallettrackr TEMP, house pattern is per-brand account → ask
a-google-accounts open, recreate under 'Sculptape' acct when David makes
it, GTM account ditto). ga4-bootstrap ran (retention/key-events/custom
dims). Runtime: products/sculptape/tracking/st-attribution.js (memo clone,
_st_eid, + NEW gtag-direct emission step — **DECISION: sculptape GA4 =
gtag-direct forever, GTM must NEVER carry GA4 tags** or double-count).
tracking_head.py stamps MO-Guard+dataLayer+gtag+runtime on all 26 pages
(pageType per page; begin_checkout / upsell_view+accept+skip Pattern H /
receipt_view wired; send_page_view:false — runtime fires page_view).
Server: netlify [functions] block + env (GA4_MEASUREMENT_ID/GA4_API_SECRET/
NEXT_WEBHOOK_SECRET/ORDER_SKU_PREFIX=ST-; secrets need explicit contexts +
scopes via accounts-API, CLI env:set 404s). **29next webhook 166 CREATED BY
API** (POST /api/admin/webhooks/ works! target /api/order-webhook,
order.created only, server-generated secret → env). **NEW TRAP+FIX:
webhooks are STORE-scoped → SS/PS orders would hit sculptape's relay →
added ORDER_SKU_PREFIX gate to shared/functions/order-webhook.mjs** (deep
SKU collect, foreign orders 200-ignored). VERIFIED: /api/capi 400-probe,
unsigned webhook 401 (HMAC armed), GA4 MP debug clean, live PDP fired
collect beacon + CAPI call, GA4 realtime showed the visit. PENDING:
a-meta-account ask (pixel+CAPI token → env FB_*, restamp heads via
builders), a-google-accounts ask (GA4/GTM per-brand accounts), then
TEST5949 E2E.
**2026-08-25 LATE-2 — META HALF LIVE: pixel 1065595576190917 "Sculptape"
CREATED BY API on Liaison–B (act_243417265464399 — David's pick, switched
from LL1 mid-turn).** SU token (adscaler FB_ACCESS_TOKEN) = the CAPI
token: proven events_received:1 w/ test code TEST5949. Env on site:
FB_PIXEL_ID + FB_CAPI_ACCESS_TOKEN(secret) + FB_TEST_EVENT_CODE=TEST5949
(REMOVE after E2E — server events land in Test Events only while set).
Pixel base (memo pattern: stub + init + noscript; runtime re-inits w/
external_id) stamped via tracking_head on funnel pages AND
homepage/collection (build_site_pages gained _with_tracking; legal pages
deliberately tracking-free). **GTM SKIPPED FOR LAUNCH (David) — GA4
gtag-direct is the permanent route; GA4 property stays under wallettrackr
acct for launch (per-brand acct = post-launch tidy).** Browser-path note:
in-app-browser resource timing can't see fbevents' sendBeacon sends —
fbq state healthy (pixel registered, userData, queue drained) + manual /tr
fetch delivered; authoritative confirm = Events Manager Test Events / stats
(<1h lag). REMAINING: TEST5949 5-phase E2E (needs a placeable order),
then drop FB_TEST_EVENT_CODE; DNS cutover re-verify.
**2026-08-25 E2E ROUND 1 (David's real order 102113, $34.98 ST-01+bump,
PAID): found + fixed 2 production bugs.** (1) Post-payment redirect 404 —
RELATIVE next-next-url resolves a level high after the processor round
trip; ENTIRE chain now root-absolute (/products/<slug>/...) — HARD RULE:
chain URLs are always absolute (HF pattern; PS's relative form was never
order-proven). (2) order-webhook 502 on every delivery — my SKU-gate
referenced out-of-scope parsed (node --check passes scope bugs!) →
scans order; HARD RULE: after ANY webhook edit, run a SIGNED synthetic
order.created probe (+ foreign-SKU probe) BEFORE any real order. Order
102113 attribution metadata was POPULATED (fbp/referrer/pixel-id
sdk 0.3.10) — no NS-trap. Redelivered signed → purchase in GA4 realtime
(2: probe + real) + Meta Test Events purchase_102113 @ TEST5949. Webhook
166 still Active. OPEN: one more full QA order to prove the ABSOLUTE
chain live (redirect → 3 OTO accepts → thank-you) + then refund 102113
(no VIP sub existed — he never saw OTOs) + drop FB_TEST_EVENT_CODE.
2026-08-25 PHASE 6 CLOSED — E2E VERIFIED on QA order 102114 ($34.98,
all OTOs DECLINED): GA4 trail proved chain (upsell_view×4 → upsell_skip×3
→ receipt_view → purchase), webhook fired ORGANICALLY, attribution full.
Meta pixel stats confirmed browser leg (PageView 9/IC 2/VC 2; sendBeacon
was invisible to resource timing only). FB_TEST_EVENT_CODE REMOVED —
production-real. **NEW ARCH RULE (capi.mjs): GA4 MP fires only for
channel:'server' events (webhook Purchase) — browser events reach GA4
via gtag only; pre-fix every event double-counted under its CAPI name
(upsell_view + UpsellView).** OPEN: refund 102113+102114; optional
accept-path order (declines proven, accepts = HF-proven SDK contract
untested live); logo pick; Sub&Save %; DNS.
2026-08-25 TBYB LANE CLOSED (last site-side gap): dedicated 4th
checkout card per David ("hide the 3 packs, show a 4th TBYB-only option")
— verified both lanes live (pkg 5 sub / pkg 6 one-time: $0.00 hydrated,
cart [tbyb,37]=$9.99, dual day-14 disclosures). Fixed en route: PDP OB-C
plan toggle rebuilt hrefs WITHOUT pkg + root-absolute (404) — now carries
PLAN_PKG ids relative; placeholder stamps must run LAST in the builder
pipeline (gate now fails on raw __TBSUB__). Site-side launch surface is
now COMPLETE; prelanders ON HOLD per David (existing dashboard owns
post-site). Remaining gates all David's: refunds ×2, logo, Sub&Save %,
DNS (then live-domain sitemap+GSC+tracking re-verify), photo shoot.
TBYB polish (David): card rebuilt on the exact template chassis
(radio + right-aligned price — never hand-build cards) and the day-14
disclosure NOTE BOXES REMOVED from all checkouts — the card pill ("$0
today · pay on day 14 only if you keep it") + Terms §3.2 carry the
disclosure; gate asserts the pill line.
2026-08-25 WAVE 7 — PDP HERO GALLERY (the 90% wave) STARTED.
Silhouette tray LIVE on the J&T board (sections 7 + 7b, module v32,
carousel_pick:true): 8 gallery candidates (hero×2, macro-peel,
in-the-box, worn-glute, application-hands, family-system, vanity-ritual)
+ 3 bundle cards (1/2/3 packs) — creative/sculptape/gallery/candidates/
silhouette/, 1024px, nano-banana i2i off w5 renders. **DAVID'S STEP-SCOPED
EXCEPTION: AI-generated imagery INCLUDING people is approved FOR THE
GALLERY WAVE ONLY (2026-08-25) — interim until photography; the global
no-AI-people rule stands everywhere else.** Selection law: click order =
slide order 01-06; thumbs 07-12 mirror slides; bundle look locks for all
zones. PARALLEL LANE OPEN: asset-sourcing run for the photographer's
per-product reference set (skills/asset-sourcing) — not yet started.
Slot-pass trap fixed same day: absolute URLs never slots (noscript pixel
beacon was slot-01, shifting everything +1).
**2026-08-25 WAVE 7 REDONE THE ADSCALER WAY (David: "hero galleries need
AdScaler, synthesized vs non-synthesized — delete the direct tray").**
Direct nano-banana tray DELETED (board v33). Dual clients LIVE on
adscalerads: **Sculptape – The Silhouette (ST-01 synthesized) 35 saved
creatives** (3.7K structured brief → Vacation Avoider/Shapewear Hater/
Skeptic personas; iMessage UGC, reverse-psychology, fake-bad-review,
testimonial formats) + (ST-01 full-brief) 39 saved (raw 07_copy-brief
response.md via catbox fytjyf.md — textarea CAPS PASTE AT 50K — →
Event-Goer/Postpartum Mom/Mature Woman personas; shapewear-vs
comparisons, feature callouts). = 74 ads. Assets: 4 catbox uploads
(o70c1d/g4t047/odzazn/dh3bp8) via DROP-event (file-input assignment
doesn't register; drag/drop DataTransfer on the zone does). **UI deltas
vs May skill: 16 format chips (not 44; hooks/copy moved to Copy tab),
chip clicks RACE React — sequential ~260ms only; profile multipanel
#conceptBrandIntelMultiList w/ data-profile-id (stale-panel rule #3
confirmed: reopen dropdown after client switch — ids must be the new
client's); board is WIDE-VIEWPORT-ONLY (narrow pane hides the whole
workspace).** NEXT: board the 74 onto the J&T ad tray (jt-ad-pipeline)
for David's verdicts — top-6 winners = gallery slides 01-06 per the
6-IMAGE-PDP-GALLERY law; then the other five zones × 2 clients each.
2026-08-25 NATIVE VERDICT BOARD LIVE: launch-center-hq.netlify.app/
sculptape/boards/sculptape--silhouette-hero-ads/ — all 74 AdScaler
creatives, pick-6 click-order = gallery slides, E2E verified + cleaned.
David picks THERE now (adscalerads boards stay as source-of-truth
archive). Verdict readback = /api/verdicts?board=… → I fill slots from
winners. SOURCING LANE OPEN: ~/.tiktok-sourcing session exists; first
pull kicked (body sculpting tape legs, the-silhouette); sourcing thumbs
→ source-approve board next (catbox-host thumbs at board build).
2026-08-25 SEQUENCING CORRECTION (David): SOURCING BEFORE GENERATION.
Sourced imagery feeds generation as anchors — the 74 existing ads stay
candidates, but the NEXT generation round (Silhouette round-2 + the five
zones) waits for sourcing approvals. LIVE: sourcing board
launch-center-hq.netlify.app/sculptape/boards/sculptape--the-silhouette-sourcing/
(20 candidates: 4 Tonee TikTok stills + 2 Skinnies + 1 creator + 5
Skinnies video frames + 8 Tonee site rips). LANES: TikTok native search
= login expired (David: pull.py --login tiktok 1 min); login-free lane
PROVEN = WebSearch URL harvest → oEmbed thumbnails → pull.py
--image-urls; yt-dlp full-video = bot-walled (latest 2026.08.19 fails;
one video snuck through). Bridge = tools/brand-hub/ingest_sourcing.py
(catbox-hosted, .catbox-cache.json dedup). FLOW: approve on board →
approved set → AdScaler Brand Assets (Lifestyle/reference) → regenerate.
**2026-08-25 CHROME SOURCING SWEEP (David's rule: TikTok = HIS Chrome via
claude-in-chrome; emulated browsers fail TikTok — no login needed there,
search renders as guest in his fingerprint).** ~70 video URLs harvested
across 5 zones via tiktok.com/search in his Chrome → oEmbed stills →
per-product ingestion. FIVE sourcing boards LIVE (per-product bridge
filters runs.json product tags; zsh no-word-split trap on $var args):
silhouette 35 · waistline 9 · lift 9 · profile 6 · neckline 7 = 66
candidates awaiting approve/hide. Arm zone query empty — rephrase later.
Flow unchanged: approvals → AdScaler assets → anchored regeneration.
2026-08-25 ANCHORED ROUND 1 LIVE (grouped source+AI board): David
approved 10 sourced images → 9 AI adaptations generated (nano i2i on the
SOURCE itself: 'recreate exact shot for SCULPTAPE, strip competitor
branding, ivory backing never blue') → grouped board
sculptape--silhouette-anchored (19 cards, 10 groups: SOURCE card then AI
VARIANT per group; board UI gained group key → section headers).
KNOWN refines: Group 4 leaked a BLUE Tonee backing; Group 1 kept a
caption ('Visible celullite' typo); one source (y-70c0229a89)
FILTER-BLOCKED → photo-team-only. Cross-zone groups 7-9 (tummy/boob/
neck adaptations) seed those zones. Winners → more variants → slides.
Netlify (2026-08-27, David): site MOVED to the EcomScalers team (teo-h3uxfci) — site id c9d5d07c-037b-43e5-99fa-1a70dcedbe05, name sculptape, https://sculptape.netlify.app. The Riptide copy (b1638592…) is DELETED. Deploy: netlify deploy --prod --dir=. --functions=../../shared/functions --site c9d5d07c-037b-43e5-99fa-1a70dcedbe05 from products/sculptape. Old site had no env vars / custom domain / aliases; nothing else to carry. RULE: new brand sites go on EcomScalers.
DOMAIN LIVE (2026-08-27): sculptape.com is attached to the EcomScalers site c9d5d07c… with SSL — https://sculptape.com/ 200, www → 301. DNS cutover is DONE (David did it after the move). pages-manifest live_urls already point at sculptape.com.
Product truth correction (David 2026-08-27): ST-04 The Neckline is a STRING-LIFT KIT like ST-05 (anchors below jaw/ear + cords toward the hairline), NOT a clear panel — copy/identity card/worn render updated; neck-specific worn render still to be made. Lift/Sleeve component renders also pending.
Neckline component LOCKED render (2026-08-27): identity/images/w6-neck-band.png — clear elastic band + white anchor pads, made in CreativeStudio from the Tonee diagram David supplied (Gemini Flash kept it nude twice; Nano Banana Pro made it truly clear — use Pro for material changes). Product truth: elastic neck-lifting band, NOT a panel, NOT strings; two tensions.
Component renders LOCKED (2026-08-27): w6-body-panel.png (Silhouette + Waistline share the SAME rounded clear panel — David), w6-arm-wrap.png (Sleeve scalloped wrap), w6-neck-band.png (Neckline band); Neckline box render w5-neckline.png now shows the band (old panel version kept as w5-neckline-panel-superseded.png, excluded from the Studio identity tab). Recipe: Nano Banana Pro "render ONLY the component from this image on the same linen, remove the box" off the w5 box render. Still pending: The Lift (nude full-cup) component.
Founder portrait = REAL photo (2026-08-28): products/sculptape/assets/pptsdr.jpg (David; same photo used for PeptideScience) baked as slot 35 default (/assets/pptsdr.webp) on all six PDPs AND wired via the Studio (founder-pptsdr.jpg identity render). AI founder shot ideas rejected. The "doctor's message" house persona now has a face — do not generate founder portraits.
Test orders to VOID: 102278 (order-summary QA replay, 2026-08-28) in addition to 102179/102180/102181/102183.
2026-09-02 late — real B/A pairs LIVE: 11 David-starred real-source pairs wired via the Studio /wire loop (Lift 29; Neckline 29/31/33; Profile 29/31/33; Silhouette 29/31; Sleeve 29/33) — CMS verified (webp 1600², ts 22:10 PDT). Waistline 29 and Sleeve 31 / Silhouette 33 still unstarred candidates. Source-picking loop + zoom law now the B/A default (see [[real-before-after-skill]]).
2026-09-03 — Waistline real B/A round: 10 real pairs as Studio candidates (29 ×4, 31 ×4, 33 ×2; strongest = kitchen creator 33, hanging belly 33, dressed side-view 31) + real-frame HIW trio on 21/22/23 (Peel/Smooth/Dress). Three TikTok rips (tummytape/apronbelly/postpartum/mumtum tags) — real-source supply is thin: ~2-3 clean befores per 14-clip rip. None starred yet.