Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Check a Canva design against a brand kit and report where it diverges — off-palette colors, non-brand fonts, logo misuse, and off-tone copy. Read-only; makes no changes. Use when the user asks \"is this on brand\", \"check this against our brand kit\", \"do a brand review\", \"does this match our brand guidelines\", or \"brand-check my design\".
.claude/skills/kunanonj-cursor-plugin-canva-brand-check/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 93% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -1% | 0% |
Compare a design against the user's brand kit and report, point by point, where it follows the brand and where it drifts. This is a specialised, brand-aware version of canva-design-feedback: same read-only "read the design, then critique" core, but the rubric is the brand kit. It never edits the design.
Canva:list-brand-kits is documented to return brand kit IDs, names, and thumbnails. It may NOT return the machine-readable palette (hex values) and font families. Your approach depends on what you actually get back:
Never invent brand colors or fonts. If you don't have the real values, say so.
Canva:get-design-content returns text only — not colors or fonts.
Canva:start-editing-transaction → inspect → Canva:cancel-editing-transaction, always cancel) reliably gives element text, positions, and sizes.Because both the brand kit (see the gap above) AND the design itself frequently lack machine-readable colors/fonts, brand-check is often a visual comparison (design thumbnail vs. brand-kit thumbnail) plus the user-supplied palette/fonts. Use Canva:get-design-thumbnail for logo placement and overall visual tone.
Short link → Canva:resolve-shortlink; full URL → extract ID; raw D... ID → use directly; otherwise ask.
Canva:list-brand-kits. If several, ask which one (or infer from team/context).Canva:get-design-thumbnail for visual/logo/tone.Check each dimension and mark On brand / Off brand / Can't verify:
Use Can't verify honestly whenever the kit didn't expose the data and the user hasn't supplied it.
## Brand check — "<design title>" vs "<brand kit name>"
Overall: ⚠️ Mostly on brand, 3 issues
### Off brand
- [Color] Page 2 heading is #1A73E8 — not in palette. Nearest brand color: #0B5CD7.
- [Font] Page 4 body uses Arial; brand body font is Inter.
### Can't verify (need input)
- Brand kit didn't expose hex values. Paste your palette + fonts for an exact check,
or confirm the colors above against your guidelines.
### On brand
- Logo placement and cover treatment match the kit.Offer to correct the API-fixable issues via canva-edit-design — note that the API can change text color and font size/weight/style, but CANNOT change font family or background colors (those are manual in Canva; see canva-edit-design).
cancel-editing-transaction after inspecting. Never commit.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 9,786 | 4,225 | -57% | 1 | 1 | 0% | 1,811 | 1,254 | -31% | 0 | 0 | — |
case-02 | fail→fail | 7,823 | 5,875 | -25% | 1 | 1 | 0% | 1,487 | 1,664 | +12% | 0 | 0 | — |
case-03 | fail→fail | 6,760 | 4,395 | -35% | 1 | 1 | 0% | 1,260 | 1,251 | -1% | 0 | 0 | — |
case-04 | fail→pass | 12,702 | 6,568 | -48% | 1 | 1 | 0% | 2,183 | 2,406 | +10% | 0 | 0 | — |
case-05 | pass→pass | 7,402 | 3,015 | -59% | 1 | 1 | 0% | 1,023 | 1,700 | +66% | 0 | 0 | — |
case-06 | fail→pass | 4,657 | 3,063 | -34% | 1 | 1 | 0% | 877 | 1,690 | +93% | 0 | 0 | — |
case-07 | pass→pass | 8,859 | 3,796 | -57% | 1 | 1 | 0% | 1,686 | 2,003 | +19% | 0 | 0 | — |
case-08 | pass→pass | 7,411 | 3,711 | -50% | 1 | 1 | 0% | 1,397 | 1,935 | +39% | 0 | 0 | — |
case-09 | pass→pass | 7,768 | 2,333 | -70% | 1 | 1 | 0% | 1,327 | 1,608 | +21% | 0 | 0 | — |
case-10 | fail→fail | 7,384 | 7,302 | -1% | 1 | 1 | 0% | 1,323 | 1,751 | +32% | 0 | 0 | — |
case-11 | fail→fail | 8,851 | 8,229 | -7% | 1 | 1 | 0% | 1,491 | 1,714 | +15% | 0 | 0 | — |
case-12 | fail→pass | 16,018 | 10,934 | -32% | 1 | 1 | 0% | 2,835 | 2,780 | -2% | 0 | 0 | — |
case-13 | pass→pass | 7,782 | 3,932 | -49% | 1 | 1 | 0% | 1,280 | 1,909 | +49% | 0 | 0 | — |
case-14 | pass→pass | 14,287 | 17,204 | +20% | 1 | 1 | 0% | 2,535 | 3,605 | +42% | 0 | 0 | — |
case-15 | fail→pass | 12,281 | 3,486 | -72% | 1 | 1 | 0% | 2,167 | 1,781 | -18% | 0 | 0 | — |
case-16 | fail→pass | 10,604 | 4,727 | -55% | 1 | 1 | 0% | 1,992 | 1,967 | -1% | 0 | 0 | — |
case-17 | pass→pass | 6,694 | 2,916 | -56% | 1 | 1 | 0% | 1,209 | 1,737 | +44% | 0 | 0 | — |
case-18 | pass→pass | 7,023 | 5,528 | -21% | 1 | 1 | 0% | 1,285 | 2,115 | +65% | 0 | 0 | — |
case-19 | fail→pass | 9,488 | 5,150 | -46% | 1 | 1 | 0% | 1,829 | 2,192 | +20% | 0 | 0 | — |
case-20 | pass→pass | 6,723 | 4,980 | -26% | 1 | 1 | 0% | 1,286 | 2,162 | +68% | 0 | 0 | — |
case-21 | pass→pass | 7,518 | 6,045 | -20% | 1 | 1 | 0% | 1,316 | 2,299 | +75% | 0 | 0 | — |
case-22 | fail→fail | 5,790 | 5,137 | -11% | 1 | 1 | 0% | 1,089 | 1,215 | +12% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 16 counted toward the lift figure. The other 6 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +27 percentage points is the difference between those two pass rates over the 16 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.