/* creator.css — THE CREATOR'S DOCUMENT GROUND (surface one; R-VS1-2/3/4, R-PAT-4, R-CGF) ITS OWN FILE ON PURPOSE. `lib/shared/tokens.css` holds the ground-set CONTRACT and no palette; keeping every declared ground out of it makes that claim true BY CONSTRUCTION rather than by a test assertion (ruling `creator-ground-file-fork`, fork (b)). Surface two gets a sibling here; `test/ground_token_contract.cjs` finds them BY PATTERN across `lib/shared/grounds/*.css`, never by a hardcoded path, so a new ground is covered the moment it lands. Values from the lab's PUBLICATION, LAB_NOTES §2026-08-28 @ `3da42023` on `lab/ui-playground`, read as a carried-out reference (R-VS1-3). The tree itself was never entered (R-LAB-3). Named `creator`, cited here, and inherited by nothing (R-VR-1). The chrome around it is untouched: the instrument stays the instrument (R-PAT-6). */ [data-doc-ground="creator"] { --doc-paper: #DEDCD4; /* pp-page — the worn page */ --doc-paper-raised: #F2F1EC; /* pp-plate — clean card stock. 1.2:1 against page is CORRECT: on paper a card separates by hairline and shadow, not tone. */ --doc-paper-sunk: #BCB7AB; /* the worst wear pixel — the darkest ground anywhere. Every ink below is solved against it, not just against the page. */ --doc-mat: #EDEBE4; /* pp-mat — the uniform image mat */ --doc-ink: #17161B; /* 13.1:1 page · 15.9:1 plate · 9.0:1 on worst wear */ --doc-ink-muted: #403D38; /* 7.9:1 · 9.5:1 · 5.4:1 */ --doc-ink-structural: #7A766D; /* 3.30:1 page — STRUCTURAL ONLY, never a readable mark */ --doc-rule: rgba(23,22,27,.28); /* R-CGG-2: the PAGE-SOLVED ochre — 6.7:1 on the page, 4.6:1 on the worst wear pixel. NOT `#826700`, which the publication solves against the PLATE only. A page-level token carrying a plate-solved value is the placement-law class in disguise: it reads correct wherever it happens to sit on card stock and fails on the page it is actually declared for. The MARK gold stays deliberately unmodeled — a fill accent, if one is ever needed, earns its own contract name at its own gate rather than being smuggled in as "the accent". */ --doc-accent: #5E4316; /* ── THE SEMANTIC FILLS — the publication's §3 and §4, entered on R-CGG-2's terms. "A fill accent earns its own contract name at its own gate, never smuggled in as the accent" — this is that gate, and these are those names. Every value below is MEASURED in the publication, not chosen here; the contrast figures are its, re-stated at the site so a later reader can check one without opening another document. THE DISTINCTION IS THE POINT, and it is the publication's own sentence: RED = a thing the character DOES — action, spend, commit, choice. INDIGO = a thing that simply IS — standing state, label, heading. On the app's near-black ground this red measured 2.3:1 and was demoted to a rule. On paper it takes its place as the headline. That is what the ground change bought, and it is why a surface may not reach for the action fill just because it wants the strong one. */ /* B-434 — THE TRUE FIGURES. This read "parchment on it 9.3:1 · it on page 7.7:1", which names ONE pair twice: contrast is SYMMETRIC, so parchment-on-red and red-on-parchment cannot differ. 7.66:1 is that pair. 9.30:1 is `--doc-paper-raised` #F2F1EC — the PLATE — on this red, i.e. a plate-solved figure written onto a token whose ink name paints PAGE parchment: the placement-law class `--doc-accent` warns about eight lines above, recurring in the next token down. Both still clear the 4.5:1 floor. */ --doc-action: #7E1414; /* parchment ink on it 7.66:1 · the plate on it 9.30:1 · it on the worst wear 5.26:1 */ --doc-action-lift: #96201C; /* HOVER ONLY — never a resting state */ --doc-action-ink: #DEDCD4; /* the page's own parchment, read back onto the fill */ --doc-state: #241E30; /* parchment on it 14.3:1 · it on page 11.7:1 */ --doc-state-ink: #DEDCD4; /* THE MARK — FILL ONLY, and the ban is the load-bearing half. 1.7:1 on the page: as type it is invisible. It was briefly the capsule default and six saturated capsules turned the quietest page in the app into a headline, so it now keeps only the job that means something — a raised characteristic, a live tick. A thing pointed AT, never a thing read. */ --doc-mark: #E0951A; --doc-mark-ink: #17161B; /* ink on gold 7.27:1 */ /* The capsule's plate (EotE p.52). `--doc-ink` on it measures 5.5:1, which is why the plate needs no ink name of its own: the document's own ink already reads on it. Its LABEL takes `--doc-state` — the maroon the page samples there measures 13.1:1 against parchment and the indigo 14.3:1, so the modeled token is the better of the two and the maroon stays deliberately unmodeled, exactly as the MARK gold was until this increment needed it. */ --doc-plate: #9C8C7C; /* ── THE STATUS INKS — the sheet's, from the books (Jason 2026-09-15, "we need status colors"; "Agree with 1": the five below). The publication modeled no status scale; the sheet's thresholds, encumbrance vital and refusals need one. These are NOT chosen: they are SAMPLED from the Beginner Game character folios (Star Wars Roleplaying/Beginner Games/Beginner Game Character Folios — Lowhhrick EotE, Cael AoR, Kaveri FaD; `Docs/SHEET_REOPENING_CENSUS.md` §4), the one colour vocabulary all three lines print the same way: the DICE, the threshold box heads, the XP chip. A player already knows what each means from the table. good = the ABILITY die's green (AoR's print, the one that clears 4.5:1 on the page) caution = the DIFFICULTY die's purple critical = the WOUNDS / STRAIN box head — the threshold red; beside --doc-action, not it live = the "CURRENT" box beside it — a FILL, like the mark: as type it is invisible spend = the XP COST chip's tan — chips and rules, not body text Proficiency yellow stays a die glyph: 1.1:1 on paper, no status can wear it. Figures: page · plate · worst wear · ink on it · parchment on it. */ --doc-good: #436F34; /* 4.29:1 · 5.21:1 · 2.95:1 · 3.05:1 · 4.29:1 */ --doc-caution: #431F4C; /* 9.94:1 · 12.07:1 · 6.82:1 · 1.32:1 · 9.94:1 */ --doc-critical: #7F2A14; /* 6.82:1 · 8.28:1 · 4.68:1 · 1.92:1 · 6.82:1 */ --doc-live: #7EA9C4; /* 1.83:1 · 2.22:1 · 1.26:1 · 7.16:1 · 1.83:1 — FILL ONLY */ --doc-spend: #A07E5A; /* 2.72:1 · 3.30:1 · 1.87:1 · 4.82:1 · 2.72:1 — chips and rules only */ /* ── THE SECTION HEADER, keyed by ground (Addendum 2, A2-2) ────────────────────────────── The approved board's sechead treatment, verified at `97226bf1`: the QUIET-LABEL register — mono face, ~12px, tracked, uppercase — in the ochre, with a thin red-tinted rule beside it. What is deliberately ABSENT is the point of the whole clause: no game-line hue (on paper the header's WORDS carry the line's identity, and the colour was only decorating them), no glow (a glow on paper is a Law 3 violation on its face, not a taste question), and no gradient wash (a printed page does not fade). These names are not part of the ground-set CONTRACT and must not become part of it: their FALLBACKS are load-bearing. `fcSectionHeaderStyle` is also called from outside any declared ground, and there the fallback IS the shipped game-line style, unchanged byte for byte until those surfaces reach their own cards. A required name would have no fallback to fall back to. */ --doc-sechead-face: var(--font-mono); --doc-sechead-weight: 400; --doc-sechead-size: 12px; --doc-sechead-track: 0.18em; --doc-sechead-ink: var(--doc-accent); --doc-sechead-shadow: none; --doc-sechead-rule: 2px solid var(--doc-action); --doc-sechead-wash: none; /* ── THE PLATE'S GEOMETRY AND MATERIAL (increment (ii), board §3) ──────────────────────── Three values the palette did not carry because they are not colours — they are how a printed card is CUT and how it sits on the page. Carried verbatim from the board. Like the sechead keys these are deliberately NOT in the ground-set contract: `paper.css` is scoped under this ground, so a surface that adopts the stylesheet declares them with its own ground or gets no plates at all — which is a loud failure, not a silent wrong one. */ --doc-cut: 12px; /* the chamfer, top-left and bottom-right */ --doc-cut-sm: 6px; /* the same cut on objects too small for 12 */ --doc-point: 14px; /* the chevron head's point depth */ --doc-unit-max: 430px; /* the phone unit's measure */ --doc-lead-max: 640px; /* the board's LEAD measure — not our prose cap */ --doc-lift: 0 1px 2px rgba(23,22,27,.16), 0 6px 16px rgba(23,22,27,.11); --doc-rule-fine: rgba(23,22,27,.17); /* the lighter hairline; .12 read as a smudge */ /* ── THE ART PLATE'S TWO, and they are DEFAULTS rather than values ──────────────────────── The species register entry sets both INLINE, per entry: the wash is that species' own portrait and the band is the frame's height. They are declared here because a name nothing defines renders UNSET, and the wash had no fallback at its use — an undefined name in a background-image makes the whole declaration invalid, which is indistinguishable from "this species has no portrait" in a screenshot. `test/doc_token_spelling.cjs` says so and is right. (Its sweep reads COMMENTS too, so this one deliberately does not spell the bare reference.) NEUTRAL BY DESIGN: `none` draws no image, which is exactly what a species without a portrait (or COMMERCIAL_MODE, where `speciesImgPath` returns null by contract) should draw. Like the sechead keys above, these are NOT part of the ground-set contract. */ --doc-wash: none; /* THE DEFAULT WASH (A1-1). `--doc-wash` is the per-species backdrop and stays `none` until one clears provenance; this is the floor under it, so a plate is never bare. It is the WEAR, and as of R-SPS-A1-1 it is the CLEAN-ROOM one at `art/wear.webp` (`wear_gen.py` @ lab `020fb64d`) rather than the CSG-derived blob this line used to name — same role, retirable upstream. Nothing new crosses to make every plate carry texture. */ --doc-wash-default: url("/art/wear.webp"); --doc-band: 340px; /* the board's own fallback at its use site */ /* THE SEVENTEEN SYMBOL INKS — the paper set, all >=4.5:1 on BOTH reading stocks (plate + mat). Token spellings are extracted from `diceUI.jsx`'s var() expressions and must match them STRING-EQUAL (R-PAT-2). The hazard is real and has THREE spellings in three documents: the publication keys it `destiny point`, the JS key is `destiny_point`, the token is `--doc-sym-destiny-point`. Only the last is correct here. */ --doc-sym-ability: #2E7931; --doc-sym-proficiency: #755D00; --doc-sym-boost: #33756D; --doc-sym-difficulty: #6B4A8F; --doc-sym-challenge: #A32626; --doc-sym-setback: #5A5A5A; --doc-sym-force: #2B2B2B; --doc-sym-success: #2D7830; --doc-sym-failure: #5A5A5A; --doc-sym-advantage: #2D7830; --doc-sym-threat: #5A5A5A; --doc-sym-triumph: #755D00; --doc-sym-despair: #A32626; --doc-sym-light: #2B2B2B; --doc-sym-dark: #6D6A62; --doc-sym-forcepip: #2B2B2B; --doc-sym-destiny-point: #7C5A24; }