Status: QUESTIONS, awaiting the partners' sitting. Ten items to settle before the look-and-feel design locks. Each carries what is already ruled or known, so nothing gets re-litigated. Companion pages: the three mockups and [the Combo](../mockups/), [the cases brief](../mockups/cases.html), [the design philosophy](../philosophy/), [the operator research](../../love-hate-research/).
Also on the table (not design): Jordan's part-tab width for MSO7 0.125" — the one number blocking all four G-code posts (-86, -87, -92, -93). Their 33199.NC measures 0.030" on flats; a word from Jordan and the programs post tonight.
The panel system: A Workbench (docked), B Studio (floating Adobe), C Graft (dock base + tear-off + workspaces). Already recommended: C built in A's order, shipped looking designed. The deciding fact is question 1.
At customer number ten: shared shop-floor PCs with rotating operators, or single-seat programmer desks? And in your own shop — does the machine operator touch cadmaster, or only whoever programs? *Shared+rotating pulls toward A; single-seat expert desks toward B; C covers both at higher cost.* Known: nothing recorded — this is a fact to gather, and Jordan's network can answer it the way his NC files answered the tab tables.
Sales, programmer, floor operator: same interface or different views? Does a floor operator get a stripped surface (their sheet, their slug report, tab-strip status) or the whole tool? Ruled already: Jobs home is the front door; the paper design carries multi-operator + per-operator audit trail. Open: who sees what.
List every panel that exists, per tab — BEFORE deciding how panels dock. The list is the real interface scope, and it is the guard against the button-pile failure: every capability must earn a home in the workflow, or it doesn't ship. Known: the mockups imply a starter set (checks, controls, parts, log/timeline, tolerance classes, to-post); nothing is ruled.
Shop lighting (is dark theme right for a bright floor?), real screen sizes, gloves/touch vs mouse, arm's-length readability. And color-blindness: the green/yellow/red status dots fail for red-green colorblind operators unless status also carries shape or position — cheap to rule now, expensive later. Known: mockup dark theme is a placeholder, explicitly.
Adobe-style shortcuts, AutoCAD-style command habits, or our own? Undo depth — and what undo MEANS when an edit re-runs gates. Known: Jordan named Adobe presets for layouts; nothing ruled for keys.
Hard blocks exist by design (missing material number, beam comp). Modal wall vs red tab vs inline flag; the wording's voice. Does in-app language follow the customer/shop vocabulary split rule? Ruled already: the gates themselves and the two-sheet split. Open: the refusal's form and tone.
Editable-everything is now principle (Jordan's story 1). Mechanics open: node editing of healed geometry? typed nudge values? When an edit re-runs the gates — instant inline, or a visible re-check step? *Jordan's daily habits should decide this one.* Ruled already: the principle itself; grabbable tabs/leads/parts in the mockups.
Which primitives (circles, rectangles, slots?), dimensioned how, and where it stops — two circles and done, or a real sketcher? Ruled already: the requirement itself (Aristide's 2" OD / 0.5" ID customer).
What a brand-new user meets: sample job? guided first heal? And how "visibly improving" shows up in-app (what's-new panel, release notes) — half of why LightBurn won (finding f1). Known: nothing ruled.
Name treatment, typography, color, icon language — and the process question: does Aristide design this layer himself, direct it, or approve it? *Design-led is the recorded philosophy; this question decides who holds the pen.* Known: philosophy recorded 2026-08-19; identity untouched.