Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when applying Anthropic brand colors/typography to an artifact: use the exact brand hex palette and Poppins/Lora type, not approximations.
.claude/skills/anthropic-brand-style/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 2 |
| gemini-3.1-pro-preview | 100% | 1 |
| Model | Lift | Δ tokens | Δ turns | Cases | Verified |
|---|---|---|---|---|---|
| gemini-3.5-flashbest | +92% | — | 0% | 24 | 86d ago |
| gemini-3.6-flash | +73% | +23% | 0% | 22 | 53d ago |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | — | — |
| case-03 | ✗→✓ | ▲ Improved | — | — |
| case-17 | ✗→✓ | ▲ Improved | — | — |
| case-02 | ✗→✓ | ▲ Improved | — | — |
| case-10 | ✗→✓ | ▲ Improved | — | — |
When an artifact must carry Anthropic's look-and-feel, render it with the EXACT brand hex values and the exact type stack below — never approximate, round, or substitute a "close enough" color or font. Apply whenever a task asks for Anthropic-branded colors, typography, or visual style.
| Token | Hex | Use for | |---|---|---| | Dark | #141413 | primary text, dark backgrounds | | Light | #faf9f5 | light backgrounds, text on dark surfaces | | Mid gray | #b0aea5 | secondary elements, muted text, borders | | Light gray | #e8e6dc | subtle/section backgrounds, callout boxes |
| Token | Hex | Rank | |---|---|---| | Orange | #d97757 | PRIMARY accent — the first accent you reach for | | Blue | #6a9bcc | secondary accent | | Green | #788c5d | tertiary accent |
#d97757 — never a generic orange like #ff6600,#ff7f50 (coral), or #e67e22.
this fixed order and repeat: orange #d97757 → blue #6a9bcc → green #788c5d → orange #d97757 → blue #6a9bcc → …
Poppins, with Arial as the fallback.Lora, with Georgia as the fallback.font-family: Poppins, Arial, sans-serif;font-family: Lora, Georgia, serif;#141413.#141413 background) is #faf9f5.#000000 or pure white #ffffff for brand text orbackgrounds — the brand neutrals #141413 / #faf9f5 replace them everywhere.
#d97757).RGBColor(0xRR, 0xGG, 0xBB) — e.g. #d97757→ RGBColor(0xD9, 0x77, 0x57), #141413 → RGBColor(0x14, 0x14, 0x13).
built-in palette entry (gray-900, orange-500).
Each BEFORE is the bare model's most common wrong default; AFTER conforms.
css/* BEFORE — generic web defaults */ .page { background: #ffffff; color: #000000; } /* AFTER — Anthropic neutrals */ .page { background: #faf9f5; color: #141413; }
css/* BEFORE — a generic "orange" */ .btn-primary { background: #ff6600; color: #ffffff; } /* AFTER — brand orange + brand light */ .btn-primary { background: #d97757; color: #faf9f5; }
css/* BEFORE — falls back to a common UI font */ h1, h2 { font-family: Helvetica, Arial, sans-serif; } /* AFTER — Poppins with declared fallback */ h1, h2 { font-family: Poppins, Arial, sans-serif; }
css/* BEFORE — a generic serif (or worse, the heading font reused) */ p { font-family: Georgia, "Times New Roman", serif; } /* AFTER — Lora with declared fallback */ p { font-family: Lora, Georgia, serif; }
css/* BEFORE — black bg, white text */ .hero { background: #000000; color: #ffffff; } /* AFTER — brand dark + brand light */ .hero { background: #141413; color: #faf9f5; }
textBEFORE: bars colored #1f77b4, #ff7f0e, #2ca02c (matplotlib defaults) AFTER: bar 1 #d97757, bar 2 #6a9bcc, bar 3 #788c5d (orange → blue → green)
css/* BEFORE — generic grays */ .border { border-color: #cccccc; } .subtle-bg { background: #f0f0f0; } /* AFTER — brand grays */ .border { border-color: #b0aea5; } /* mid gray */ .subtle-bg { background: #e8e6dc; } /* light gray */
python# BEFORE — raw black + a guessed orange shape.fill.fore_color.rgb = RGBColor(0, 0, 0) # AFTER — brand dark, exact channels shape.fill.fore_color.rgb = RGBColor(0x14, 0x14, 0x13) accent.fill.fore_color.rgb = RGBColor(0xD9, 0x77, 0x57) # #d97757
#d97757. It is always the first pick.#d97757,card 5 is blue #6a9bcc, and so on. Do not invent a fourth accent.
#faf9f5 text on orange/green/bluefills (not #ffffff), and #141413 if the design calls for dark text.
gray #e8e6dc, distinct from the page background #faf9f5.
headings, Georgia for body. Keep Poppins/Lora first in the stack regardless.
#6a9bcc,tertiary is green #788c5d — never a tint of the orange.
#000000 / #ffffff. Always #141413 / #faf9f5.#ff6600, #ff7f50). Always #d97757.Lora (body).
Poppins, Arial, … and Lora, Georgia, ….#FFF, #D97757). Always 6-digit lowercase.#ffffff/#000000 because they are the web defaults.#d97757.of the orange→blue→green cycle.
#ccc, #999) for #b0aea5 / #e8e6dc.#faf9f5 and #141413, never #fff/#000.#d97757.#d97757 → blue #6a9bcc → green #788c5d.#b0aea5 (mid) and #e8e6dc (light).| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.5-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 +73 percentage points is the difference between those two pass rates over the 22 comparable cases.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.5-flash | verified | 6/27/2026 | +92% |
Other measured skills in the registry, with their headline benchmark lift.