NOTES

What still needs a decision

This page was produced end to end by an unattended Claude session on 31 August 2026, running the eleven-stage pipeline in ~/Programming/single-page-sales against the product in ~/Programming/navy-diving. Nothing on it has been approved by Tim. The list below is everything a human has to settle before it could go anywhere near the public internet, ordered by how badly it blocks.

Blocking

  1. DECISION · LARGEST ASSUMPTION ON THE PAGE Is this for sale at all? Nothing in the navy-diving repo says it is a product. The whole page rests on assuming it is. If the answer is no, delete the page and keep the argument, because the argument is good and it belongs in the README.
  2. DECISION The prices are invented. $49 and $129 were chosen by the session to have something concrete to write against. There is no pricing anywhere in the source.
  3. DECISION Selling to service members. Study material for military qualification boards, priced, sold to E-4s. It is legal and the market for it exists, but it is a positioning call and no one has made it.
  4. BLOCKER Nobody has read the 965 cards. The review viewer at navy-diving-cards.pages.dev is built and live and the verdict table is empty. The page says this out loud, twice, in its largest disclosure section. Selling a safety-adjacent deck that its own author has not read end to end is a real risk even with the disclosure. Recommendation: review at minimum the 195 judgment cards and the 116 treatment-table cards before any money changes hands.
  5. BLOCKER There is no checkout, no delivery, and no license file. Every CTA points at /get-started, which says so plainly. The offer describes an email with files in it. No such email exists.
  6. DECISION The dive locker license is undefined. The page says "one command's locker." That is not a license. Who counts, for how long, what happens at a transfer.
  7. DECISION The guarantee is a proposal. Full refund, any time, no questions. Nobody has agreed to honour that.
  8. DECISION A full facilitator packet is published free at /sample. That is one of the fourteen drills, given away in its entirety, as the page's proof asset. It tested as the single strongest conversion element. It is still a giveaway that has not been approved.
  9. CHECK The Anki trademark. The page names Anki throughout, describes AnkiApp and AlgoApp accurately as unaffiliated, and states that the deck is not affiliated with the Anki project. That reads as ordinary descriptive use, but nobody has actually checked it against Anki's own guidance.

Review, not blocking

  • The gauntlet was self-run. Stage 08 requires the evaluator to be a separate context from the drafter. This session had subagents ruled out, so the ten personas were simulated in-context. Scores (4.46 average, 9 of 10 would buy) are a structured self-critique, not evidence about real readers.
  • No real diver has read this page. The single highest-value next step is showing it to one Master Diver for twenty minutes.
  • Doc drift found in navy-diving. CLAUDE.md still lists topic::seamanship as "nearly empty." The built report says 78 cards after the Salvor's Handbook pass. The docs are behind the build.
  • Every claim was re-derived this session. The gate was run, the citation verifier was run, and card counts were tallied from the YAML rather than read out of a README. The log is at pages/usn-diving-master/07-evidence/output/evidence-log.md. Four claims failed verification during that pass and were corrected or cut.
  • Deliberately absent: urgency, scarcity, countdowns, testimonials, trust badges, pass rates, and any claim that a person has reviewed the card content. None of those are oversights.

Where everything lives

  • Pipeline instance: ~/Programming/single-page-sales/pages/usn-diving-master/
  • Final copy: 11-polish/output/final-sales-page.md
  • Evidence log: 07-evidence/output/evidence-log.md
  • Gauntlet: 08-test-readers/output/feedback-synthesis.md
  • Scorecard: 11-polish/output/quality-scorecard.md
  • The product itself: ~/Programming/navy-diving/