jetthoughts · pftg · Aug 21, 2026 · Aug 21, 2026
diff --git a/.okf/design/landing-anatomy.md b/.okf/design/landing-anatomy.md
@@ -0,0 +1,37 @@
+---
+type: design-principle
+title: Landing anatomy is settled; register is the variable
+description: Service/sales pages share one landing skeleton; design comparisons vary only the visual register on top of it
+tags: [design, landing-page, redesign, v2]
+timestamp: 2026-08-21T00:00:00Z
+---
+
+# Landing anatomy is settled; register is the variable
+
+A design **system** is not a page **design**. Applying good tokens to article
+anatomy produced a page Paul read as "a blog, not a presentation" (PR-2,
+2026-08-21). Sales pages are designed as landing pages; the reusable system is
+extracted from shipped pages afterwards — never the reverse.
+
+## The settled skeleton (three independent sources converged)
+
+Problem-naming hero (+ dual CTA + de-risk line + stat strip) → month-one card →
+decision table (vs the alternatives) → free-audit-first timeline → case cards →
+named verified testimonial (dark band) → closing CTA. Convergence: Paul's
+Claude Design artifact `3dc819cf`, the 20.05 spec, and the 2026 external sweep
+(`docs/projects/2608-site-design-system/10-19-research/10.01-design-register-research-2026-08-reference.md`).
+
+## Rules
+
+- Pilots/comparisons vary the **register** (palette, type, surface language)
+  on the shared skeleton — never anatomy and skin at once, or the comparison
+  measures nothing.
+- Proof sits near decision points, not in a basement section; CTAs repeat at
+  multiple scroll points; forms ≤5 fields.
+- Every number on the page passes [/content/claims-canon.md](/content/claims-canon.md)
+  BEFORE styling work starts; tenure derives from `site.Params.foundingYear`.
+
+# Citations
+
+- Research + candidate shortlist: `docs/projects/2608-site-design-system/10-19-research/10.01-design-register-research-2026-08-reference.md`
+- Plan of record: `docs/projects/2608-site-design-system/20-29-strategy/20.06-clean-slate-v2-plan.md` §Flow revision
diff --git a/.okf/log.md b/.okf/log.md
@@ -51,6 +51,20 @@ make it green: restructure same-day entries under one heading, and add
 `timestamp` to the 23 concepts missing it (anchored to each file's last commit
 time, which is verifiable - never invented).
 
+## 2026-08-21 - anatomy settled, register under test: the pilots flow
+
+Paul's PR-2 review ("looks like a blog, not a presentation") exposed that we
+styled article anatomy instead of designing a landing page - a design system
+is not a page design. His finished Claude Design artifact (`3dc819cf`)
+independently produced the same landing skeleton as the 20.05 spec and the
+external research sweep, so the anatomy question is closed and the register
+(palette/type/surface) is the only variable worth piloting. New concept:
+[landing-anatomy](design/landing-anatomy.md); research + 3-candidate shortlist:
+`docs/projects/2608-site-design-system/10-19-research/10.01-*.md`. Flow is now
+research -> 2-3 register pilots on /next/ (shared skeleton) -> lightning demo.
+Claims gate before styling: the CD artifact carries pricing/audit-promise/
+stay-stats NOT in canon - Paul approves or they come out.
+
 ## 2026-08-21 - PR-2 pilot: two CSS traps and a claims fix, from adversarial review
 
 The styled D1 pilot shipped through author -> verifier -> fixes. Three durable

diff --git a/...esign-system/10-19-research/10.01-design-register-research-2026-08-reference.md b/...esign-system/10-19-research/10.01-design-register-research-2026-08-reference.md
@@ -0,0 +1,77 @@
+# Design-register research → pilot candidates (2026-08-21)
+
+Directive (Paul, 2026-08-21): Claude Design + Stitch are collaborative
+surfaces, not one-shot generators. Flow: **online research → top 2–3 design
+systems → one pilot each on the `/next/` rail → lightning demo → decide
+together.** This doc is the research leg and the candidate shortlist.
+
+## The finding that reframes the work
+
+Paul's finished Claude Design artifact (project `3dc819cf-084f-4dd6-b7aa-7275c2424e68`,
+"Fractional CTO.dc.html") independently converged on the SAME landing anatomy
+as our unstarted 20.05 spec: problem-naming hero, month-one card,
+fractional-vs-full-time decision table, free-audit-first timeline, case cards,
+named verified testimonial, repeated CTA. External research (below) converges
+on it too.
+
+**Conclusion: the anatomy is settled. The variable worth piloting is the
+REGISTER** (palette + type + surface language). All pilots share one skeleton;
+each wears a different skin. That makes the lightning demo a controlled
+comparison instead of apples-to-oranges.
+
+## Research notes (what the sweep supports)
+
+- **Trust cues are the differentiator for skeptical B2B buyers**: third-party
+  review ratings, named quotes with real title/company, quantified outcomes,
+  case studies — placed above the fold and near decision points, not in a
+  basement section. Buyers protect their own career; sloppy design reads as
+  risk. (trajectorywebdesign, userevidence, windmillstrategy, bopdesign)
+- **More information converts for complex/high-ticket B2B** — comparison
+  tables, FAQs, process transparency; CTAs repeat at multiple scroll points;
+  forms ≤5 fields. (genpage, unicornplatform, perspective)
+- **2026 register split**: (a) *editorial text-first* — typography as the
+  interface, serif boldness, page-weight discipline; (b) *engineered
+  precision* — sharp geometry, 1px borders, stark neo-grotesque, "premium
+  brutalism". Both are anti-template signals; decoration is what reads
+  template. (designmonks, setproduct, fireart, brainy.ink)
+- **Peer register (thoughtbot.com)**: spacious dark-first sans, quantified
+  proof ("1,000+ positive outcomes"), conversational CTAs ("Let's talk"),
+  generous whitespace = confidence. Validates restraint-over-decoration for
+  dev consultancies.
+
+## Candidate registers (all share the settled anatomy)
+
+| | A — Rescue Room (warm humanist) | B — Artifact Editorial | C — Engineered Precision |
+|---|---|---|---|
+| Source | Paul's finished CD artifact `3dc819cf` | D1, panel winner 4–1; Paul's CD pass in progress `8eb98dd6` | Research (2026 premium-brutalist) + D2's authority intent, re-skinned light |
+| Surfaces | warm paper `#faf7f3`/`#f2ece4`, white cards, 16px radius, dark closing band `#14110f` | paper `#fbf9f8`, 0px corners, ruby hairlines | white/near-white, 1px structural borders, sharp geometry, generous white |
+| Type | Poppins headings + system body | Source Serif 4 + Hanken Grotesk | stark neo-grotesque (e.g. Inter/Söhne-class) + mono accents |
+| Says to a burned founder | "safe, human, we'll take care of you" | "senior, published, thoughtful" | "precise, engineered, nothing hidden" |
+| Risk | rounded warmth can read generic-SaaS | can read blog/essay if anatomy slips (the PR-2 lesson) | can read cold/intimidating to non-technical buyers |
+| Ruby #cc342d | primary CTA | hairlines + CTA | restrained accent only |
+
+## Claims gate (BLOCKING before any pilot ships)
+
+From the CD artifact, verified against `.okf/content/claims-canon.md`:
+- ✓ canon: 4.8/5 Clutch (linked), avg client relationship 5 years, developer
+  experience 8+ avg, Bruno Wozniak/PubNative = the single real testimonial.
+- ✗ NOT in canon — Paul must approve or they come out: **$5K–$15K/month**
+  pricing, **free code audit** + **five working days**, **30 days' notice**,
+  **"most clients stay 3+ years / some beyond 6"**.
+- "18+" is derived from `site.Params.foundingYear` in templates, never frozen.
+
+## Pilot mechanics
+
+One stub per candidate on the rail: `/next/pilots/<register>/fractional-cto/`
+(cascade noindex covers all of `/next/`), each with its own CSS bundle —
+the rail's per-page CSS ownership makes a 3-way skin comparison ~free.
+Lightning demo = side-by-side links + screenshots + 3-line verdicts per lens
+(5-second trust, CTA clarity, mobile), then discussion; Claude Design and
+Stitch remain the iteration surfaces for the chosen register.
+
+Sources: sitebuilderreport/knapsackcreative/colorlib (consulting examples);
+cleardigital/bopdesign/windmillstrategy/lowcode (B2B 2026 trends);
+trajectorywebdesign/userevidence/hginsights (skeptical-buyer trust);
+designmonks/setproduct/fireart/brainy.ink (typography + premium-brutalism);
+genpage/unicornplatform/perspective/popupsmart (landing anatomy);
+thoughtbot.com (peer register, fetched 2026-08-21).
diff --git a/docs/projects/2608-site-design-system/20-29-strategy/20.06-clean-slate-v2-plan.md b/docs/projects/2608-site-design-system/20-29-strategy/20.06-clean-slate-v2-plan.md
@@ -32,6 +32,24 @@ a non-indexed `/next/` path on production for live testing.
 | Legacy | Frozen (defect fixes only). 780 button sites + 39 blacks DROPPED as work-done-twice. |
 | Sub-agent models | sonnet/haiku; Fable reserved for main loop + critical judgment |
 
+### Flow revision (Paul, 2026-08-21, after PR-2 review)
+
+Two corrections superseded the single-winner competition:
+1. **"Looks like a blog, not a presentation"** — the pilot applied D1 tokens to
+   article anatomy. A design *system* is not a page *design*; pages are
+   designed as landing pages, the system is extracted from shipped pages.
+2. **Compare pilots, don't crown early** — new flow: online research → top 2–3
+   register candidates → **one pilot per candidate on `/next/`, all sharing one
+   landing anatomy** → lightning demo, decide together. Claude Design + Stitch
+   are the collaborative iteration surfaces throughout.
+
+Research + shortlist (A Rescue Room warm-humanist / B Artifact Editorial /
+C Engineered Precision) + the BLOCKING claims gate:
+`../10-19-research/10.01-design-register-research-2026-08-reference.md`.
+The anatomy is settled (Paul's Claude Design artifact `3dc819cf`, the 20.05
+skeleton, and external research all converged on it); registers are the
+variable under test.
+
 ## The two tracks
 
 ### Track A — design collaboration (no repo code; parallel with any PR)
@@ -131,11 +149,17 @@ same commit · browser-measured AA + seams on anything touching a rendered page.
 | PR-0: 2.3 course eyebrow | ALL GATES DONE on PR #552: opus review (4 findings fixed), merge-base attribution (delta = exactly 2 course keys), Linux baselines screened 2-of-72, codex findings applied | merge #552; **OUTSTANDING: macOS desktop+mobile course/chapter baselines need recording on a NON-drifted Mac** — this machine proved 13-red on untouched master |
 | Track A: direction 2 (stitch) | DONE — "Institutional Authority", screen `7cbde8e2e694438d82138be4d6f7f3f9`; canon flags in 20.07 |
 | Track A: Claude Design | project created via DesignSync (`a5f25203-6dd9-45c6-a978-5842152d1389`), seeded with brief + both directions | Paul iterates there; then judging |
-| Track A: judging board | VOTE RUN — **D1 wins 4–1** (tally + per-voter reasoning in 20.07); D1 = working winner, Paul's Claude Design pass can override; /next/ styling fully reversible | Paul confirms or overrides in project `a5f25203` |
+| Track A: judging board | VOTE RUN — **D1 wins 4–1** (20.07) — then SUPERSEDED by the register-pilots flow (§Flow revision): no early crown, compare pilots instead | — |
+| Register research + shortlist | DONE — `10-19-research/10.01`: anatomy settled, 3 register candidates (A Rescue Room / B Artifact Editorial / C Engineered Precision), claims gate listed | — |
+| Paul's Claude Design artifacts | A finished (`3dc819cf`, full landing page — the anatomy blueprint); B in progress (`8eb98dd6`) | pilots port from these; B refreshes when Paul finishes |
+| Pilot A (Rescue Room) | NOT STARTED | port CD artifact `3dc819cf` to `/next/pilots/rescue-room/fractional-cto/`; claims gate FIRST (pricing/audit-promise/stay-stats need Paul) |
+| Pilot B (Artifact Editorial) | NOT STARTED — existing `/next/services/fractional-cto/` has the skin but article anatomy | rebuild on the shared landing skeleton |
+| Pilot C (Engineered Precision) | NOT STARTED | tokens from 10.01 research; same skeleton |
+| Lightning demo | BLOCKED on pilots | side-by-side links + screenshots + 3-line verdicts per lens; discussion; Paul picks |
 | PR-1: v2 rail | **MERGED** #555 (`703fee12c`) — rail live: `/next/` noindexed staging, qtest holes closed (root layouts + v2 registry-derived keys), ADR-0006, mermaid source-aware. Screenshot job red = the documented 16-key #540 handoff only, verified key-by-key. | — |
 | PR-2: styled pilot | **MERGED** #557 (`38f744805`) — full content ported (overview/faqs/tenure-derived stats), styled per D1, canon-exact Clutch proof line, browser-verified AA (button 5.13:1), −85.6% gzip CSS vs legacy page. Screenshot red = the same 16-key #540 handoff set, verified key-by-key. | compare links: `/next/services/fractional-cto/` vs `/services/fractional-cto/` |
-| PR-3: flip | BLOCKED on Paul's live review of the pilot (+ his design-pass verdict) | flip = `type: next` on the real page + delete stub; checklist in ADR-0006 §5 |
-| 20.05 service-page spec | NOT STARTED (hive specialists idle) | content model (month-one card, decision table, timeline) survives any design winner |
+| PR-3: flip | BLOCKED on lightning-demo winner + Paul's live review | flip = `type: next` on the real page + delete stub; checklist in ADR-0006 §5 |
+| 20.05 service-page spec | VALIDATED by convergence — Paul's CD artifact independently produced the same skeleton (10.01 §finding) | the shared pilot anatomy IS this spec; no separate doc needed |
 | Homepage read gate | WAITING ≈2026-09-17 | first: confirm `e1fa5409d` deploy date |
 
 **Standing rules for whoever picks this up:** WIP=1 on repo PRs; sub-agents on