Own the AryaOS Cockpit theme (v1.2.0)

[ampledata]

Eleven cockpit-* plugins each carried their own 143–308 line stylesheet defining an identical --aos-* palette. cockpit-dronecot and cockpit-lincot were byte-for-byte identical; charontak was a 308-line variant of the same thing; four more repos also shared an identical 230-line pkg/lib/page.scss.

None referenced the AryaOS Product Suite design tokens, so the console had drifted from the kit and from aryaos.org:

	Cockpit shipped	Kit says
accent	#35a58f teal	Signal Orange #E4610F
ok	#74d8a4	#3F7D57
danger	#ff8a7a	#A32820
body font	RedHatText	Archivo
mono	ui-monospace	JetBrains Mono
cards	8px radius + 0 10px 30px shadow	radius 0, no shadows

What the plugins already had right, and this preserves: the dark console theme, 48px (3rem) glove targets, and one 760px breakpoint with no strays anywhere. That part of the fleet sweep held.

Aliases, not a rewrite

--aos-* names are kept as aliases onto the kit tokens. Every plugin rule already says var(--aos-panel), so aliasing makes those rules kit-correct without rewriting hundreds of declarations. A plugin migrates by deleting its palette block and importing this.

The fonts are bundled, not just named

Cockpit renders in the operator's browser, not on the Pi — so installing Archivo on the box would do nothing. Measured on aryaos-4f11: fc-list finds 0 Archivo and 0 JetBrains faces. A bare font-family: Archivo would have silently fallen back to sans-serif: a rule that looks applied and isn't.

Per-plugin webfonts are already the established pattern here — PatternFly ships ~22 woff2 into every plugin and @font-faces them relatively (264 across the fleet). Ours add ~85 KB. OFL-1.1 licence and notices travel with them.

Two documented deviations

1. --sns-target: 48px, not the kit's 44px — operator surfaces are used with gloves.
2. Status colours lightened for the console, because the kit specifies them against Paper. Measured WCAG contrast of the kit values as text on #0B1512:

ok    #3F7D57  3.79:1  passes UI 3:1, FAILS body text 4.5:1
warn  #A8811C  5.15:1  passes -- would have been fine as-is
fault #A32820  2.55:1  FAILS outright

The hue stays anchored to the kit token and the console variant is derived with color-mix, so there's no second hardcoded palette to drift. Percentages put all three in one narrow band, because a triad lightened unevenly reads as mismatched: ok 65% → #7EA589 6.75:1, warn 75% → #BA9C4F 7.04:1, fault 50% → #CA8C84 6.72:1.

Scoping of radius-0 / no-shadow

Scoped to our own surfaces plus the PatternFly components we actually render — not a blanket * rule. PF uses radius meaningfully in switches, spinners and progress bars. .aos-dot keeps its 50% and says why: a 2px square status dot reads as a defect, not a state.

Verified

Compiles to 4839 bytes. Asserted on the compiled output: radius 0, target 48px, #e4610f, Archivo, JetBrains Mono, box-shadow: none, 760px and the woff2 URL all present — and the only non-zero radius emitted is the documented dot. Tests 6 passed, typecheck clean.

Next

This is the foundation only. The 11 plugin migrations (delete palette block, import theme, drop the shadow/radius rules) follow separately so each can be eyeballed on .13 rather than landing 11 repos blind.

🤖 Generated with Claude Code

https://claude.ai/code/session_0197da7dhvcPoHxYKamrYqyM