AetherCutChangelogHow it worksPricingOpen the editor

Home / Changelog / v1.9.91

v1.9.91

Share buttons, Ctrl+J mode cycle, pricing tripwire, adoption KPI

AetherCut v1.9.91 shipped on Aug 13, 2026.

What's new in v1.9.91?

  • Share buttons on /six-tools — one-tap X/Twitter + LinkedIn share intents + Copy Link button. Shared URLs carry ?utm=share so recipient clicks hit the share-arrival counter shipped in v1.9.90. Only rendered on /six-tools (not /vs/* pages) so the offensive positioning stays the primary share surface.
  • Ctrl+J / Cmd+J shortcut — cycles the sidebar mode picker (all → media → cut → style → all). Symmetric with Ctrl+K (CommandPalette). Toast confirms every switch. Routes through handleSetSidebarMode so keyboard-driven switches also populate the mode-distribution analytics.
  • Mode Adoption KPI card — new triple StatCard in AdminFunnel above the per-mode distribution row: 'adoption %' (filtered / total), 'total mode selections', 'filtered non-all selections'. Interprets the % into 'load-bearing / earns its space / mostly decorative' so the on-call product answer is instant.
  • Pricing snapshot regression — new pricingSnapshot.regression.test.mjs asserts every $X.XX/mo mention in seo_landings.py maps to a canonical tier, that the /six-tools OG generator + copy references AI Layer ($9.99/mo) + Full Pro ($14.95/mo) NOT the $4.99 Suite, and that PRO_PRICE_NOTE covers monthly/yearly/lifetime. This is the tripwire — pricing has drifted twice (v1.9.10 shipped $9/mo instead of $14.95/mo; v1.9.90 first shipped $4.99 for AI flagships)
  • Regression suite grew to 422 (+9 new: 6 pricing snapshot, 3 v1.9.91 next-actions)

How to validate the release without fooling yourself

The thread running through v1.9.91 is not “more UI,” it’s tighter attribution. The share buttons only matter if they create traffic you can recognize, the sidebar shortcut only matters if keyboard usage lands in the same analytics path as clicks, and the pricing tripwire only matters if it catches copy drift before a page quietly starts selling the wrong plan. That makes this a release that benefits from a short, opinionated validation pass instead of broad smoke testing. A practical way to do it is to treat each change as a chain with a visible user action on one end and a measurable internal consequence on the other. On /six-tools, the action is tapping X, LinkedIn, or Copy Link; the consequence is that the outgoing URL includes ?utm=share and a recipient visit increments the share-arrival counter introduced in v1.9.90. For the mode cycle shortcut, the action is pressing Ctrl+J or Cmd+J repeatedly; the consequence is not just that the sidebar changes and a toast appears, but that the switch flows through handleSetSidebarMode so the distribution data remains comparable across mouse and keyboard behavior. For pricing, the action is less interactive but more dangerous: editing any landing copy that mentions a monthly price. The consequence should be deterministic test failure the moment someone references a non-canonical value or surfaces the wrong tier on /six-tools.

One useful release-day drill is to simulate a complete share loop with a clean browser session instead of checking only the rendered buttons. Open /six-tools, trigger each share affordance, and inspect the produced URL rather than assuming the intent target preserved the query string. Then paste that exact copied URL into a separate browser profile or private window and confirm the arrival path behaves like a genuine recipient click. This catches a subtle class of false positives where the origin page renders everything correctly, but downstream handling strips the parameter or a copy helper grabs the canonical URL without the tracking suffix. The same “full loop” mindset applies to the shortcut: cycle through all modes several times and verify the order stays all → media → cut → style → all, because a single stale enum or picker refactor can keep the toast honest while shifting the actual analytics labels underneath. A common mistake here is testing only from the default all state once, which won’t expose an off-by-one bug or a mode that becomes unreachable after repeated keypresses.

The new KPI card is especially easy to misread if you look at it before you decide what question it answers. It is built for the on-call product question, not for a full strategy review. “Adoption %” is a filtered-versus-total ratio, while the companion counts give the denominator context: total mode selections and filtered non-all selections. In practice, that means the card is strongest when you’re trying to distinguish “nobody uses this because nobody sees it” from “people see it and still don’t choose it.” If filtered non-all selections are moving but adoption remains weak, the mode likely exists in users’ awareness yet fails the value test. If total selections rise while filtered selections stay flat, users may be cycling through the surface without finding a reason to leave all. The built-in labels — “load-bearing,” “earns its space,” “mostly decorative” — are useful because they compress triage time, but they should not replace checking whether a recent launch, documentation update, or keyboard shortcut changed the composition of who is selecting modes in the first place.

The pricing regression deserves the strictest discipline because it protects trust, not just consistency. Pricing drift is rarely introduced by a “pricing change”; it usually sneaks in through marketing edits, OG copy tweaks, or a well-meaning landing-page update that repeats an old number from memory. The new test is valuable precisely because it treats every $X.XX/mo mention as suspicious until proven canonical, and because it explicitly guards the high-intent /six-tools surface from advertising the cheaper Suite when the intended references are AI Layer at $9.99/mo and Full Pro at $14.95/mo. The practical habit to adopt is simple: whenever you touch seo_landings.py, /six-tools share copy, or anything that reuses pricing notes, run the regression before you review the prose. It is faster to fix a failing mapping than to reason manually about whether a sentence “sounds right,” and it prevents the most expensive kind of copy bug: one that looks polished, ships cleanly, and teaches users the wrong price.

How do I get this update?

Open the editor and you have it. AetherCut runs in Chrome, Edge, Safari and Firefox. There is nothing to install, and you can edit without an account. Open the editor or see how it works.

What shipped just before and after?

See all 111 releases · Follow the RSS feed