CtrlK
BlogDocsLog inGet started
Tessl Logo

adobe/stardust

Redesign an existing website to make it better. Built on top of impeccable.

71

Quality

89%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Low

Low-risk findings worth noting

Overview
Quality
Evals
Security
Files

composition-delta-good.mdskills/prototype/fixtures/


slug: beers url: https://wasatchback.com/beers register: brand surprise: high dominantDimension: ground-family/label-substrate-zine

Page shape: beers (variant C) — composition-delta-good fixture

Worked example of a shape brief whose compositionDelta blocks declare ≥ 2 structural changes per pair against both siblings. This fixture passes Discipline 10 (variant-convergence detector) under ia-fidelity: reimagined.

Sections (in render order)

  1. eyebrow-banner (site-wide chrome)
  2. header (site-wide chrome)
  3. page-hero — direction-authorized type-led intro
  4. catalog-color-block × 6 — per-SKU color-keyed zine spreads
  5. catalog-closer-on-black — return to brand-black anchor
  6. footer (site-wide chrome)

CompositionDelta — pairwise structural deltas

_provenance.compositionDelta_vs_A:
  - "substrate-strategy: A is single black across all; C uses 6 distinct per-section substrates (one per beer's label color)"
  - "section-presence: C drops filter-chip-row (substrate sequence IS the navigation cue); also drops SEASONAL divider"
  - "photo-treatment: A photo at 35% column with hairline thumbnail; C photo full-bleed-within-section as label-art-foreground"
  - "section-order: A hero → filter → list → closer; C hero → 6 color blocks → closer-on-black (return to ground)"
  - "section-completeness: A entries are content blocks on shared ground; C entries are their own visual environments (each block IS its own zine spread)"

_provenance.compositionDelta_vs_B:
  - "substrate: B is uniform black; C is per-block per-substrate"
  - "photo-role: B is small 180px framed thumbnail; C is full-bleed within section column"
  - "page-rhythm: B is typographic parade (names at clamp(5rem, 12vw, 8rem)); C is zine color-spread (names at clamp(3.5rem, 8vw, 6rem); the substrate carries impact, not size)"
  - "type-scale on name: B bigger (clamp 5-8rem); C smaller (clamp 3.5-6rem) — different captured-trait amplification (photography wallpaper-color vs names-as-design)"

Why this passes Discipline 10

Pair (A, C) has 5 structural deltas — substrate-strategy, section-presence, photo-treatment, section-order, section- completeness. The minimum is 2.

Pair (B, C) has 4 structural deltas — substrate, photo-role, page-rhythm, type-scale. The minimum is 2.

All deltas are structural (composition, sequence, presence, layout-strategy, page-rhythm) rather than token-level ("variant uses a different teal", "variant uses font weight 600 vs 400"). Token-only deltas would fail the validator with "trivial compositionDelta — token-level only; ideation restart required."

Each variant has a distinct dominantDimension:

  • A: composition/catalog-grid (template-shaped, faithful)
  • B: typography/name-as-display (vernacular-naming amplified)
  • C: ground-family/label-substrate-zine (label-color-as-design amplified)

The three dimensions are non-overlapping; no variant re-uses another's dimension.

How Discipline 10 reads this

The validator runs pairwise on the brief set {A-shape.md, B-shape.md, C-shape.md}. For each pair, it:

  1. Reads _provenance.compositionDelta_vs_<sibling>[].
  2. Filters deltas to structural changes (regex-classifies each delta's leading token: substrate*, section-*, photo-*, layout*, ia-priority*, page-rhythm*, type-scale* count as structural; pure color, font-weight, font-family, motion-energy, spacing-rhythm count as surface-only).
  3. Requires ≥ 2 structural deltas under ia-fidelity: reimagined; rejects when the count drops below.

This fixture's vs-A pair has 5 structural deltas; vs-B has 4. Both clear the floor.

skills

README.md

tile.json