Redesign an existing website to make it better. Built on top of impeccable.
71
89%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Low
Low-risk findings worth noting
Worked example of a shape brief whose compositionDelta declares
only token-level deltas. This fixture fails Discipline 10
under ia-fidelity: reimagined and triggers an ideation restart.
_provenance.compositionDelta_vs_A:
- "color: A uses teal accent #1c8a7a; C uses warmer teal #1a8870"
- "font-weight: A uses 600 on headlines; C uses 700"
- "spacing-rhythm: A uses 32px section gap; C uses 40px"
_provenance.compositionDelta_vs_B:
- "color: B uses orange accent #e8b830; C uses warmer teal #1a8870"
- "motion-energy: B has 240ms hover; C has 320ms hover"All deltas in both pairs are surface-only (token-level): color hex tweaks, font-weight bumps, spacing changes, motion- energy adjustments. None of them are structural (composition, sequence, presence, layout-strategy, page-rhythm, type-scale, ia-priority).
The validator classifies each delta's leading token:
| Delta | Classification |
|---|---|
color: A uses teal accent ... | surface (color) |
font-weight: A uses 600 ... | surface (font-weight) |
spacing-rhythm: A uses 32px ... | surface (spacing-rhythm) |
color: B uses orange ... | surface (color) |
motion-energy: B has 240ms ... | surface (motion-energy) |
Under ia-fidelity: reimagined, the validator requires
≥ 2 structural deltas per pair. This fixture has 0
structural deltas on pair (A, C) and 0 on pair (B, C).
The validator refuses with:
Discipline 10 failure: variant C declares trivial compositionDelta — token-level only against both siblings. Variant convergence detected (per
notes/variant-convergence.md); the variant is a token reskin of variant A. Restart ideation:
- amplify a different captured trait than B (currently both variants amplify the same composition)
- declare structural moves (substrate / section / layout / IA priority / page-rhythm / type-scale) against the captured cliché
- re-author the brief with
compositionDeltablocks carrying ≥ 2 structural deltas per pair
The Discipline 10 validator's test suite runs this fixture expecting refusal:
const result = validateDiscipline10(briefs);
assert.equal(result.status, "refused");
assert.match(result.reason, /trivial compositionDelta/);The good fixture (composition-delta-good.md) is the accept
case in the same suite.
Under ia-fidelity: verbatim, this fixture's surface-only deltas
would PASS Discipline 10's inverse contract (A1/A2/A3 must
differ on surface axes only). The validator inverts:
reimagined → structural ≥ 2, surface-only failsverbatim → surface ≥ 2, structural failsThis file is named composition-delta-trivial.md because the
deltas are trivial for the reimagined contract — they are also
the canonical good form for the verbatim contract. The
verbatim test fixture would re-use the same delta block,
re-stamped as A2 / A3 against A1.
.tessl-plugin
skills
audit
reference
deploy
diff
direct
extract
migrate
prepare-migration
prototype
replica
reskin
rollout
stardust
uplift
reference