Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create a content style guide / voice & tone guide so everyone writes consistently. Use when asked to write a content style guide, a voice and tone guide, editorial guidelines, or UX-writing standards. Produces a usable guide — voice principles with do/don't examples, tone-by-context, mechanics (grammar, capitalisation, formatting), terminology/word list, and accessibility/inclusivity rules — that a team can actually apply.
.claude/skills/mohitagw15856-content-style-guide/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 28% | 0% |
A style guide makes a brand sound like one voice no matter who's writing. The useful ones aren't 50 pages of rules — they're voice principles with examples, tone guidance by context, and a word list people reach for daily. This skill produces a guide a team will actually use, grounded in concrete do/don't examples rather than abstract adjectives.
Given "a style guide for our fintech app", produce the full guide anyway — infer voice principles and terminology from the brand and audience, and mark inferred choices for the team to confirm. Make every principle show an example. Never hand back abstract values with no examples.
Ask for these only if they aren't already provided (else infer and label):
1. Voice — who we are — 3–4 voice principles, each as "We are X, not Y" with a before/after example.
2. Tone — how we adapt — how the voice flexes by context (e.g. celebratory on success, calm and brief on errors, warm in onboarding), with a small table: situation → tone → example.
3. Mechanics — the rules that come up constantly: capitalisation (sentence vs. title case), punctuation (Oxford comma, exclamation marks), numbers/dates/currency, contractions, US/UK spelling, formatting (headings, lists, links, buttons).
4. Word list — a do/don't terminology table: preferred term, what to avoid, and why (product terms, jargon to drop, words that are on/off-brand).
5. Inclusivity & accessibility — inclusive language, reading level, plain-language rules, and accessibility (link text, alt text, no "click here", no directional-only instructions).
6. Quick reference — a one-screen cheat sheet of the most-used rules.
Mark inferred voice/terminology choices (confirm with the team).
Content design practice — example-driven voice principles, context-based tone, editorial mechanics, terminology management, and inclusive/accessible language.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 93,865 | 24,727 | -74% | 1 | 1 | 0% | 5,242 | 4,977 | -5% | 0 | 0 | — |
case-02 | fail→pass | 97,839 | 42,166 | -57% | 1 | 1 | 0% | 5,749 | 5,231 | -9% | 0 | 0 | — |
case-03 | pass→fail | 7,874 | 5,474 | -30% | 1 | 1 | 0% | 1,469 | 1,772 | +21% | 0 | 0 | — |
case-04 | pass→pass | 12,276 | 16,313 | +33% | 1 | 1 | 0% | 1,938 | 3,310 | +71% | 0 | 0 | — |
case-05 | pass→pass | 17,128 | 24,551 | +43% | 1 | 1 | 0% | 3,823 | 4,549 | +19% | 0 | 0 | — |
case-06 | fail→pass | 20,253 | 24,619 | +22% | 1 | 1 | 0% | 3,222 | 4,738 | +47% | 0 | 0 | — |
case-07 | pass→pass | 18,180 | 23,289 | +28% | 1 | 1 | 0% | 2,879 | 4,573 | +59% | 0 | 0 | — |
case-08 | pass→pass | 25,344 | 23,576 | -7% | 1 | 1 | 0% | 4,152 | 4,590 | +11% | 0 | 0 | — |
case-09 | fail→pass | 21,545 | 24,312 | +13% | 1 | 1 | 0% | 3,554 | 4,797 | +35% | 0 | 0 | — |
case-10 | pass→pass | 17,564 | 23,891 | +36% | 1 | 1 | 0% | 2,818 | 4,441 | +58% | 0 | 0 | — |
case-11 | fail→pass | 25,345 | 25,009 | -1% | 1 | 1 | 0% | 3,872 | 4,944 | +28% | 0 | 0 | — |
case-12 | pass→pass | 78,024 | 25,438 | -67% | 1 | 1 | 0% | 2,750 | 4,638 | +69% | 0 | 0 | — |
case-13 | fail→pass | 85,144 | 26,086 | -69% | 1 | 1 | 0% | 3,424 | 5,055 | +48% | 0 | 0 | — |
case-14 | fail→pass | 15,411 | 15,970 | +4% | 1 | 1 | 0% | 2,538 | 3,604 | +42% | 0 | 0 | — |
case-15 | pass→pass | 17,852 | 17,969 | +1% | 1 | 1 | 0% | 2,934 | 4,090 | +39% | 0 | 0 | — |
case-16 | pass→pass | 25,727 | 24,790 | -4% | 1 | 1 | 0% | 2,871 | 4,725 | +65% | 0 | 0 | — |
case-17 | fail→pass | 17,722 | 23,610 | +33% | 1 | 1 | 0% | 3,265 | 4,534 | +39% | 0 | 0 | — |
case-18 | fail→fail | 18,086 | 19,506 | +8% | 1 | 1 | 0% | 3,091 | 4,287 | +39% | 0 | 0 | — |
case-19 | fail→pass | 25,139 | 22,997 | -9% | 1 | 1 | 0% | 4,352 | 4,547 | +4% | 0 | 0 | — |
case-20 | pass→pass | 16,474 | 20,233 | +23% | 1 | 1 | 0% | 2,650 | 4,355 | +64% | 0 | 0 | — |
case-21 | pass→pass | 17,000 | 24,103 | +42% | 1 | 1 | 0% | 2,914 | 5,011 | +72% | 0 | 0 | — |
case-22 | pass→pass | 21,167 | 24,979 | +18% | 1 | 1 | 0% | 3,561 | 5,036 | +41% | 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 +36 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.