sheleg-design-skill 1.32.0 → 1.33.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -4,6 +4,55 @@ All notable changes to this project are documented in this file. The format
4
4
  follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); versions
5
5
  follow [SemVer](https://semver.org/spec/v2.0.0.html).
6
6
 
7
+ ## [1.33.0] - 2026-08-15
8
+
9
+ **Three things the pack asserts that a running page disproved.** Same dashboard as 1.32.0,
10
+ audited a second time — but by walking the **computed styles of every element** rather
11
+ than by re-reading the source. That is the difference this entry is about: all three
12
+ findings are invisible to a grep over the stylesheets, and two of them are invisible to
13
+ the stylesheets entirely.
14
+
15
+ ### Fixed — in `tenor`
16
+
17
+ - **The second-hue ban was being broken by the pack's own token, on every render.** Bans
18
+ said "the orange is the entire chromatic budget", flatly, while the Palette section
19
+ three hundred lines above derived `--good` and explained why a product surface needs it.
20
+ Measured on a live screen, hue 150 appeared **36 times on one page** — checklist marks
21
+ and success chips. The ban now names its exception and states the real rule: no hue you
22
+ did not find in this file, and the only two in it are the orange and `--good`.
23
+ - **Two bans the browser breaks for you, with no stylesheet involved.** `<strong>` renders
24
+ at **700** against "no weight above 500" and `<em>` renders italic against "no italics".
25
+ Neither appears in any CSS, so neither is greppable and neither surfaces in review;
26
+ both were found by reading computed styles. `tokens/tenor.css` now ships the three-line
27
+ base layer, and emphasis resolves the pack's own way — by **value**, not by weight or
28
+ slant. **Every pack in this library that bans a weight or a slant owes the same block**,
29
+ and none of the twenty ships one; filed as `B-044`.
30
+ - **The 25px orange rule conflated a measurement with a limit.** It is the size of the
31
+ reference's investor lockup, not a threshold derived from anything, and a product's own
32
+ lockup will not be 25px. The rule it stands for is *one small resting orange object per
33
+ screen*; the entry now says which half is which.
34
+
35
+ ### Added — in `tenor`
36
+
37
+ - **`--data-rest`**, a role over the value ramp the staircase already spends — `--stair-3`
38
+ on the paper (3.37:1 on `--bg`, over the non-text floor) and `--ink-faint` on the band.
39
+ It exists because this pack has **one border weight**: a design system carrying
40
+ `--border` and `--border-strong` as two steps collapses both onto `--line` here, and a
41
+ chart that reached for the stronger of the two paints its resting bars in the exact
42
+ colour of the rule between its own rows. Measured on a dashboard's account table — the
43
+ two resolved identically and the bars matched the row separator to the byte. Nothing
44
+ errors, because both names resolve; that is the failure mode.
45
+ - The `### On a product surface` section gains the chart rule, and repeats the standing
46
+ instruction that charts in any pack go through `dataviz` first.
47
+
48
+ ### Note
49
+
50
+ Two contrast claims in this diff named `--line`, which is declared with alpha and is
51
+ therefore dropped from the validator's solids — so no claim naming it can ever be checked.
52
+ `validate_stated_ratios` caught the one that sat on a checkable line and stayed silent on
53
+ the one that did not, which is the hole `B-003` already describes. Both are now written as
54
+ arguments with their resolved hex rather than as measurements.
55
+
7
56
  ## [1.32.0] - 2026-08-15
8
57
 
9
58
  **The pack met a product surface, and three of its own measurements did not survive
@@ -108,6 +108,20 @@
108
108
  --stair-2: oklch(89% 0 0);
109
109
  --stair-3: oklch(62% 0 0);
110
110
 
111
+ /* A resting data mark — a sparkline bar, an unemphasised column, a track.
112
+ A role name over the value ramp above rather than a new value, on the same
113
+ pattern as --sev-*. It exists because this pack has ONE border weight, so a
114
+ system carrying `--border` and `--border-strong` as two steps collapses both
115
+ onto --line here, and a chart that reached for the stronger of the two ends
116
+ up painting a quantity in the colour of the rule between its own rows.
117
+ --stair-3 resolves to #868686, and
118
+ --data-rest is 3.37:1 on --bg — over the 3:1 non-text floor for a mark.
119
+ Against the hairline it is 2.38, computed on the resolved #d2d1ce and NOT
120
+ checkable from here: --line is declared with alpha, so the validator drops
121
+ it from the solids and no claim naming it can be verified. Stated as an
122
+ argument rather than as a measurement, deliberately. */
123
+ --data-rest: var(--stair-3);
124
+
111
125
  /* ── geometry: there is no radius and there are no shadows ───────────────
112
126
  Zero occurrences of border-radius and zero of box-shadow in the whole
113
127
  stylesheet. Separation is the hairline; emphasis is a filled rectangle. */
@@ -231,6 +245,12 @@
231
245
  /* --accent-ink is deliberately NOT overridden: the orange is the same orange
232
246
  on both fields, so the one label that can be read on it is the same coal. */
233
247
  --sev-limit: #4a4a48;
248
+ /* The value ramp is a paper-page device and inverts to nothing useful here, so
249
+ the resting mark takes the band's own faint ink instead, where
250
+ --data-rest is 4.69:1 on --bg. Against the hairline it is 2.64 on the
251
+ resolved #3e3e3c — same caveat as the light block: --line carries alpha and
252
+ cannot be checked from here. */
253
+ --data-rest: var(--ink-faint);
234
254
  /* DERIVED, same reasoning as the light block and re-measured on this field:
235
255
  8.45:1 on --bg, and 10.7 from the accent under deuteranopia. A surface
236
256
  block that remaps eleven tokens and forgets the status set is a defect this
@@ -239,6 +259,25 @@
239
259
  --danger: #f7f6f2;
240
260
  }
241
261
 
262
+ /* The base layer, and it is three lines because the browser breaks two of this
263
+ pack's bans on its own. `<strong>` renders at 700 against "no weight above
264
+ 500", and `<em>` renders italic against "no italics" — with no stylesheet
265
+ involved either time, so neither violation is greppable and neither appears in
266
+ review. Found by walking the computed styles of a running page rather than the
267
+ source.
268
+
269
+ Emphasis resolves the pack's own way, by VALUE: both take the full ink against
270
+ prose set in --ink-soft-aa, and `<strong>` stops at the pack's ceiling weight.
271
+ Any pack in this library that bans a weight or a slant owes the same block. */
272
+ strong, b {
273
+ font-weight: var(--weight-mono);
274
+ color: var(--ink);
275
+ }
276
+ em, i {
277
+ font-style: normal;
278
+ color: var(--ink);
279
+ }
280
+
242
281
  /* Reduced motion. The reference collapses every duration to 0.01ms with
243
282
  iteration-count clamped to 1, which lands an animation on its final frame
244
283
  rather than strobing it, and then names its one infinite loop separately to
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "sheleg-design-skill",
3
- "version": "1.32.0",
3
+ "version": "1.33.0",
4
4
  "description": "Design taste as an installable agent skill. Cinematic scroll-driven landing pages built on one scroll clock and layered degrade-to-calm motion, a motion doctrine that decides whether to animate before it decides how, three calibration dials, and twenty locked style packs with ready-made design tokens — instrument-console, editorial-luxury, workbench, briefing-room, atrium, orchard, field-notes, cyclorama, showroom, blueprint, prism, maquette, scoreboard, datasheet, manpage, pigeonhole, roster, ora, tenor and paperclip. Colour, slop and fork-reciprocity gates run as scripts, not opinions. Works with Cursor, Claude Code and any agent that reads a SKILL.md.",
5
5
  "bin": {
6
6
  "sheleg-design-skill": "bin/cli.js"
@@ -2,7 +2,7 @@
2
2
  "name": "sheleg-design",
3
3
  "displayName": "SHELEG Design",
4
4
  "description": "SHELEG Design methodology: cinematic scroll-driven landing pages (single scroll clock, layered degrade-to-calm motion, WebGL particle formations), a motion doctrine that decides whether to animate before it decides how, and twenty pluggable visual style packs — instrument-console (dark console), editorial-luxury (warm editorial), workbench (light/dark product UI for dashboards and tools), briefing-room (dark 16:9 deck), atrium (warm consumer health), orchard (friendly consumer biotech), field-notes (warm paper for dev tools sold on auditability), cyclorama (a pastel field on a 32s cycle), showroom (the product as the exhibit), blueprint (a drawing sheet, zero radius), prism (one iridescent wash over mono body), maquette (cream axonometric models on a dark table), scoreboard (warm paper, pixel numerals, a dark ledger of results), datasheet (an off-white spec sheet whose live instrument goes dark when it detects the reader is hiding), manpage (a developer landing page set in the reader's own system monospace on cream paper, with coral label chips that are real headings and a dark code frame as the argument), pigeonhole (a white sorting wall whose nine pastel categories are a filing scheme rather than a mood, each a two-layer chip whose label word is mandatory), roster (a white field in a faint grid of squares whose argument is other people's marks — client logotypes in pill-labelled industry columns, an engine's wordmark inside the headline — where the proof is a name rather than a number), ora (a warm coal field with cream ink and no third hue — the accent is the inverted field — a serif carrying every human sentence over a monospace carrying every machine fact, a terminal surface cut below the page, and a six-step verdict ramp, for products whose output is a machine's verdict about the reader), tenor (warm paper with zero radius and zero shadow, one hairline weight, an orange that exists only on hover and on focus, a sans tracked negative against a mono tracked positive, display at a line-height below one in an eight-to-twelve-character measure, and product proof delivered as silent looping video, for products arguing that a new kind of thing must be managed like an existing organisation), paperclip (neutral coal with no functional colour at all — every control monochrome and elevation made of hairlines, with the whole chromatic budget spent on a curtain of gradient capsules and twelve gradient section badges that cannot be clicked — for products that ask a person to run something that runs itself: agent teams, autonomous back-office, schedulers and budget-governed compute). Ships the sheleg-design skill, the architecture reference, the motion doctrine, the Figma and Claude Design bridges, AI-surface patterns, style packs with ready-made token CSS, and the /sheleg-design command.",
5
- "version": "1.32.0",
5
+ "version": "1.33.0",
6
6
  "author": {
7
7
  "name": "ssheleg",
8
8
  "url": "https://x.com/sshlg93"
@@ -3,7 +3,7 @@ name: sheleg-design
3
3
  description: Use when building or upgrading a cinematic scroll-driven landing page, marketing site or hero (particle/WebGL background, scroll-linked animation, parallax, scrubbed sections) — when such a page feels busy or janky or its motion layers drift out of sync — or when styling product UI with its style packs - dashboards, admin panels, internal/dev tools, mobile app screens, design tokens, light/dark themes - or when carrying a visual system across the Figma border (publishing tokens as variables, implementing a design without importing raw values). Triggers - "cinematic landing" / "кинематографичный лендинг", "scroll animation" / "скролл-анимация", "dashboard style" / "стиль дашборда", "design tokens" / "дизайн-токены", "light/dark theme" / "светлая/тёмная тема", "figma variables / figma to code" / "переменные фигмы, фигма в код", "chat/agent UI" / "интерфейс чата или агента", "streaming output" / "стриминг ответа", "mobile screen" / "мобильный экран".
4
4
  license: MIT
5
5
  metadata:
6
- version: 1.32.0
6
+ version: 1.33.0
7
7
  ---
8
8
 
9
9
  # SHELEG Design
@@ -299,6 +299,21 @@ mounting the pack on a populated twenty-route dashboard rather than reasoned abo
299
299
  across a dashboard's stat row reads exactly as it does on the page.
300
300
  - **Prose sits on two fields at once and `--ink-soft` fails on the second.** This
301
301
  is what produced `--ink-soft-aa`; see Gotcha 1.
302
+ - **A data mark must not borrow the hairline, and in this pack it silently will.**
303
+ Tenor has one border weight and no strong variant, so a design system that
304
+ carries `--border` and `--border-strong` as two steps collapses them onto one
305
+ value here. A sparkline whose resting bars were painted with the *stronger* of
306
+ those two is then painted in the exact colour of the rule between the rows it
307
+ sits in — a quantity rendered in the token reserved for chrome, and nothing
308
+ errors, because both names resolve. Measured on a dashboard's account table:
309
+ the two resolved identically and the bars matched the row separator to the
310
+ byte. **Use `--data-rest`**, which the token layer maps onto the value ramp the
311
+ staircase already spends: `--stair-3` on the paper (3.37:1 on `--bg`, and 2.38
312
+ against `--line`, so it separates from the rule) and `--ink-faint` on the band.
313
+ The emphasised mark stays `--accent-ink`-on-`--ink` or whatever the chart's own
314
+ emphasis rule says; only the resting one changes. Charts in any pack go through
315
+ `dataviz` first — token names are not uniform across the twenty, and an aliased
316
+ one fails exactly this quietly.
302
317
 
303
318
  What does **not** change on a product surface: the spacing scale and the type
304
319
  ramp. Tenor's section rhythm is `clamp(40px, 4.5vw, 64px)` and its display slope
@@ -457,11 +472,35 @@ hierarchy is spent.
457
472
  drawing.
458
473
  - **No shadows, and no elevation model at all.** Something is either filled or it
459
474
  is not. A card with a shadow is a different pack.
460
- - **No second hue.** The orange is the entire chromatic budget, and it is spent on
461
- hover and focus. A resting orange fill larger than 25px is out of register.
475
+ - **No second hue — with one named exception, and it is `--good`.** The orange is
476
+ the chromatic budget of the *page*, spent on hover and focus. The pack also
477
+ ships a derived success value at hue 150 for the product surfaces it dresses,
478
+ because a marketing site paints no success state and a dashboard has to. So the
479
+ ban as it stands is: **no hue you did not find in this file**, and the only two
480
+ in it are the orange and `--good`. A screen may carry both; nothing else.
481
+
482
+ *This used to read "the orange is the entire chromatic budget", flatly, while
483
+ the Palette section three hundred lines above derived `--good` and said why.
484
+ Measured on a dashboard running this pack, hue 150 appeared **36 times on one
485
+ screen** — checklist marks and success chips — so the ban was being broken by
486
+ the pack's own token on every render.*
487
+
488
+ A resting orange fill larger than 25px is out of register. **That number is a
489
+ measurement of the reference's investor lockup, not a limit derived from
490
+ anything** — a product's own lockup will not be 25px, and the rule it stands
491
+ for is *one small resting orange object per screen*. Treat the size as the
492
+ reference's and the count as the rule.
462
493
  - **No weight above 500, and no weight above 400 in the sans.** Emphasis comes
463
494
  from size, measure and fill.
464
495
  - **No italics.** The reference contains none.
496
+ - **Both of the two above need a base layer, because the browser breaks them for
497
+ you.** `<strong>` renders at **700** and `<em>` renders italic with no
498
+ stylesheet involved, so neither violation appears in any CSS a grep can read
499
+ and neither shows up in review. `tokens/tenor.css` ships the reset; emphasis
500
+ resolves the pack's own way, by **value** — `<strong>` takes weight 500 and the
501
+ full `--ink` against prose set in `--ink-soft-aa`, and `<em>` drops the slant
502
+ and takes the ink too. Any pack in this library that bans a weight or a slant
503
+ owes the same three lines.
465
504
  - **No gradient with a hue in it.** The one gradient in the system is an
466
505
  achromatic greyscale wash behind a single feature cell.
467
506
  - **No centred display type**, except the hero below 620px and the closing panel.
@@ -108,6 +108,20 @@
108
108
  --stair-2: oklch(89% 0 0);
109
109
  --stair-3: oklch(62% 0 0);
110
110
 
111
+ /* A resting data mark — a sparkline bar, an unemphasised column, a track.
112
+ A role name over the value ramp above rather than a new value, on the same
113
+ pattern as --sev-*. It exists because this pack has ONE border weight, so a
114
+ system carrying `--border` and `--border-strong` as two steps collapses both
115
+ onto --line here, and a chart that reached for the stronger of the two ends
116
+ up painting a quantity in the colour of the rule between its own rows.
117
+ --stair-3 resolves to #868686, and
118
+ --data-rest is 3.37:1 on --bg — over the 3:1 non-text floor for a mark.
119
+ Against the hairline it is 2.38, computed on the resolved #d2d1ce and NOT
120
+ checkable from here: --line is declared with alpha, so the validator drops
121
+ it from the solids and no claim naming it can be verified. Stated as an
122
+ argument rather than as a measurement, deliberately. */
123
+ --data-rest: var(--stair-3);
124
+
111
125
  /* ── geometry: there is no radius and there are no shadows ───────────────
112
126
  Zero occurrences of border-radius and zero of box-shadow in the whole
113
127
  stylesheet. Separation is the hairline; emphasis is a filled rectangle. */
@@ -231,6 +245,12 @@
231
245
  /* --accent-ink is deliberately NOT overridden: the orange is the same orange
232
246
  on both fields, so the one label that can be read on it is the same coal. */
233
247
  --sev-limit: #4a4a48;
248
+ /* The value ramp is a paper-page device and inverts to nothing useful here, so
249
+ the resting mark takes the band's own faint ink instead, where
250
+ --data-rest is 4.69:1 on --bg. Against the hairline it is 2.64 on the
251
+ resolved #3e3e3c — same caveat as the light block: --line carries alpha and
252
+ cannot be checked from here. */
253
+ --data-rest: var(--ink-faint);
234
254
  /* DERIVED, same reasoning as the light block and re-measured on this field:
235
255
  8.45:1 on --bg, and 10.7 from the accent under deuteranopia. A surface
236
256
  block that remaps eleven tokens and forgets the status set is a defect this
@@ -239,6 +259,25 @@
239
259
  --danger: #f7f6f2;
240
260
  }
241
261
 
262
+ /* The base layer, and it is three lines because the browser breaks two of this
263
+ pack's bans on its own. `<strong>` renders at 700 against "no weight above
264
+ 500", and `<em>` renders italic against "no italics" — with no stylesheet
265
+ involved either time, so neither violation is greppable and neither appears in
266
+ review. Found by walking the computed styles of a running page rather than the
267
+ source.
268
+
269
+ Emphasis resolves the pack's own way, by VALUE: both take the full ink against
270
+ prose set in --ink-soft-aa, and `<strong>` stops at the pack's ceiling weight.
271
+ Any pack in this library that bans a weight or a slant owes the same block. */
272
+ strong, b {
273
+ font-weight: var(--weight-mono);
274
+ color: var(--ink);
275
+ }
276
+ em, i {
277
+ font-style: normal;
278
+ color: var(--ink);
279
+ }
280
+
242
281
  /* Reduced motion. The reference collapses every duration to 0.01ms with
243
282
  iteration-count clamped to 1, which lands an animation on its final frame
244
283
  rather than strobing it, and then names its one infinite loop separately to