Validation

Software evidence

Executed October2, 2026, using Node24.14.1 and the production build at /Package-Forge/ on port4191:

Domain/storage coverage includes independent $96/$960/$80 Studio and $2,880/$240 Archive examples, free viewers, zero/large/negative/decimal counts, month/year/leap boundaries, no-fit, missing approvals, exact downgrade losses, immutable quote snapshots, stale drafts/selections/revisions, quote arithmetic and references, disappearing schedules, superseded/canceled distinction, unexpected assumptions, expired quotes/stale catalog and getter/read/write failures.

Production coverage includes Cedar preview/cancel, annual commitment and refresh, Archive upgrade, Base downgrade and exact losses, replacement/cancel preservation, schedule cancellation without revival, Widefield no-fit, separately reviewed must-have changes, scoped Undo, invalid and zero counts, compatible refresh, corrupt/unknown/inconsistent storage preservation and explicit reset, same-revision content change, cross-tab corruption, getter/read/write failures, trap Tab/reverse Tab/Escape, focus restoration and fallback, 1280×633 editor/quote visibility, 320/390 mobile layouts, docs HTTP200, no page/console errors and no external app requests. Additional regressions verify that unreadable storage cannot overwrite unseen original bytes, stale reset cannot erase another tab's newer commitment, failed reset preserves original corrupt bytes, and expired/catalog records are explained and preserved.

Real screenshots at 1280×633, 320×844 and 390×844 were inspected. Review actions stay visible beneath scrollable details; narrow screens have no horizontal overflow. Screenshots and generated reports remain ignored local evidence rather than public product data. Terminal color warnings are tooling-only; no application error was observed. Browser/preview processes owned by implementation were stopped after verification.

React review checked stable dialog effects, derived fit/cost state, explicit button actions, accessible labels/fieldset/announcements, dialog keyboard boundaries and focus fallback. The app uses no data fetch, third-party font or external runtime service. These software checks do not establish human comprehension or real billing reliability. Local storage has no atomic multi-tab transaction; a race after the checked snapshot remains a prototype limit.

Proposed human evaluation — not performed

Ask a participant to choose Cedar's lowest-cost fitting tier, explain annual charge versus equivalent monthly, and state Kite's exact lost capability/capacities and effective date. Observe correct package selection, total-cost explanation and downgrade consequence comprehension without hints. No participant results, customer/commercial outcomes or owner comprehension have been observed. These proposed measures do not establish willingness to pay.

Release evidence

Verified October 2, 2026.

The primary reviewer independently checked the fixed Studio baseline, reviewed an eight-editor annual quote of $960 with $80 equivalent monthly, compared the $2,880 Archive quote, canceled and confirmed a replacement, reviewed the $0 Base downgrade with explicit lost capabilities/capacity, canceled its schedule without reviving an older one, and verified the no-fit scenario. Three immutable quote records and cancellation survived refresh. Scenario entry focused the editor at 1280×633; the 320px page had no horizontal overflow or reported page/console errors.

Initial release d828249579baf683c87f60c242244224f533720e passed GitHub verification and Pages deployment. Local HEAD matched public main, the worktree was clean, and all 10 deployed files matched the local production build and the GitHub deployment artifact byte for byte. The live page rendered its expected entry controls without reported page/console errors. The first deployment attempt reached Pages before hosting was enabled and returned404; verification had passed. After workflow hosting was enabled, only the failed deployment job was rerun successfully. No application change was needed for that setup timing issue.

The public profile links the case study, PRD, walkthrough and live demo. The assistant delivery and publication work is complete for this bounded prototype. Final documentation revisions repeat the verification/publication pipeline; the private delivery ledger records final-head parity. These are point-in-time software observations, not uptime or commercial-outcome claims.

Mo's personal comprehension and endorsement of provisional choices, and actual human evaluation, remain unobserved. Software checks cannot establish those findings. This is the publication-status record referenced by the other product documents.