Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Provides the single source of truth for brand guidelines, design tokens, technology choices, and voice/tone. Use this skill whenever generating UI components, styling applications, writing copy, or creating user-facing assets to ensure brand consistency.
.claude/skills/valec3-maintaining-brand-identity/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 3% | 0% |
Brand Name: INSERT BRAND NAME HERE]
This skill defines the core constraints for visual design and technical implementation for the brand. You must adhere to these guidelines strictly to maintain consistency.
Depending on the task you are performing, consult the specific resource files below. Do not guess brand elements; always read the corresponding file.
If you need exact colors, fonts, border radii, or spacing values, read: 👉 resources/design-tokens.json
If you are generating code, choosing libraries, or structuring UI components, read the technical constraints here: 👉 resources/tech-stack.md
If you are writing marketing copy, error messages, documentation, or user-facing text, read the persona guidelines here: 👉 resources/voice-tone.md
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 11,882 | 19,377 | +63% | 1 | 1 | 0% | 2,707 | 4,502 | +66% | 0 | 0 | — |
case-02 | fail→fail | 12,412 | 11,784 | -5% | 1 | 1 | 0% | 1,777 | 2,292 | +29% | 0 | 0 | — |
case-03 | fail→pass | 14,443 | 12,369 | -14% | 1 | 1 | 0% | 3,269 | 3,022 | -8% | 0 | 0 | — |
case-04 | fail→fail | 9,965 | 9,310 | -7% | 1 | 1 | 0% | 1,434 | 1,797 | +25% | 0 | 0 | — |
case-05 | fail→pass | 10,893 | 6,313 | -42% | 1 | 1 | 0% | 2,050 | 1,197 | -42% | 0 | 0 | — |
case-06 | fail→pass | 11,127 | 12,387 | +11% | 1 | 1 | 0% | 2,341 | 2,901 | +24% | 0 | 0 | — |
case-07 | fail→pass | 13,846 | 12,126 | -12% | 1 | 1 | 0% | 2,236 | 2,208 | -1% | 0 | 0 | — |
case-08 | fail→fail | 17,284 | 20,598 | +19% | 1 | 1 | 0% | 3,745 | 4,323 | +15% | 0 | 0 | — |
case-09 | fail→pass | 5,440 | 4,855 | -11% | 1 | 1 | 0% | 1,025 | 1,057 | +3% | 0 | 0 | — |
case-10 | fail→fail | 11,954 | 9,320 | -22% | 1 | 1 | 0% | 1,964 | 1,920 | -2% | 0 | 0 | — |
case-11 | fail→pass | 4,338 | 7,116 | +64% | 1 | 1 | 0% | 638 | 1,175 | +84% | 0 | 0 | — |
case-12 | fail→pass | 12,933 | 11,736 | -9% | 1 | 1 | 0% | 2,426 | 2,644 | +9% | 0 | 0 | — |
case-13 | fail→pass | 9,223 | 8,013 | -13% | 1 | 1 | 0% | 1,535 | 1,440 | -6% | 0 | 0 | — |
case-14 | fail→fail | 18,756 | 16,171 | -14% | 1 | 1 | 0% | 4,109 | 3,360 | -18% | 0 | 0 | — |
case-15 | fail→pass | 28,096 | 12,752 | -55% | 1 | 1 | 0% | 4,003 | 3,265 | -18% | 0 | 0 | — |
case-16 | fail→fail | 9,257 | 8,260 | -11% | 1 | 1 | 0% | 1,546 | 1,687 | +9% | 0 | 0 | — |
case-17 | fail→fail | 14,698 | 15,145 | +3% | 1 | 1 | 0% | 3,247 | 3,765 | +16% | 0 | 0 | — |
case-18 | fail→pass | 5,721 | 4,306 | -25% | 1 | 1 | 0% | 899 | 1,073 | +19% | 0 | 0 | — |
case-19 | fail→pass | 9,808 | 9,311 | -5% | 1 | 1 | 0% | 1,700 | 1,637 | -4% | 0 | 0 | — |
case-20 | pass→pass | 4,368 | 4,490 | +3% | 1 | 1 | 0% | 930 | 1,070 | +15% | 0 | 0 | — |
case-21 | pass→pass | 6,922 | 6,337 | -8% | 1 | 1 | 0% | 1,440 | 1,431 | -1% | 0 | 0 | — |
case-22 | pass→pass | 8,989 | 7,051 | -22% | 1 | 1 | 0% | 1,772 | 1,595 | -10% | 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. The headline lift of +50 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.