Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write concise Diataxis docs for gh-aw with Starlight markdown conventions.
.claude/skills/github-documentation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 118% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 91% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 109% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 69% | 0% |
Documentation lives in docs/, uses GitHub-flavored Markdown, renders with Astro Starlight, and follows Diátaxis.
Organize documentation into four Diátaxis types:
Purpose: Guide beginners through achieving a specific outcome to build confidence.
Avoid: Explaining concepts in depth, multiple options, troubleshooting
Purpose: Show how to solve a specific real-world problem or accomplish a particular task.
Avoid: Teaching fundamentals, explaining every detail, being exhaustive
Purpose: Provide accurate, complete technical descriptions of the system.
Avoid: Instructions, tutorials, opinions on usage
Purpose: Clarify and illuminate topics to deepen understanding.
Avoid: Step-by-step instructions, exhaustive reference material
aw for agentic workflow snippets with YAML frontmatterExample workflow code block:
aw wrapon: push # Your workflow steps here
Documentation files use GitHub-flavored markdown with Astro Starlight for rendering. Key syntax elements:
Every documentation page must have frontmatter:
markdowntitle: Page Title description: Brief description for SEO and navigation
Use GitHub's alert syntax for notes, tips, warnings, and cautions:
markdown> [!NOTE] > Important information the reader should notice. > [!TIP] > Helpful advice for the reader. > [!WARNING] > Warning about potential issues or pitfalls. > [!CAUTION] > Critical warning about dangerous operations. > [!IMPORTANT] > Key information users need to know.
title attribute for file names: yaml title=".github/workflows/example.yml" aw language for agentic workflow files with YAML frontmatterwrap for line wrapping: aw wrap Use tabs for showing alternatives (e.g., different languages, platforms):
markdownimport { Tabs, TabItem } from '@astrojs/starlight/components'; <Tabs> <TabItem label="npm"> ```bash npm install package ``` </TabItem> <TabItem label="yarn"> ```bash yarn add package ``` </TabItem> </Tabs>
Use cards for navigation or highlighting multiple options:
markdownimport { Card, CardGrid } from '@astrojs/starlight/components'; <CardGrid> <Card title="Getting Started" icon="rocket"> Quick introduction to the basics. </Card> <Card title="Advanced Usage" icon="setting"> Deep dive into advanced features. </Card> </CardGrid>
Remember: Keep components minimal. Prefer standard markdown when possible.
Documentation bloat reduces clarity and makes content harder to navigate. Common types of bloat include:
When editing documentation, focus on:
Consolidate bullet points:
Eliminate duplicates:
Condense verbose text:
Standardize structure:
Simplify code samples:
Before (Bloated):
markdown### Tool Name Description of the tool. - **What it does**: This tool does X, Y, and Z - **Why it's valuable**: It's valuable because A, B, and C - **How to use**: You use it by doing steps 1, 2, 3, 4, 5 - **When to use**: Use it when you need X - **Benefits**: Gets you benefit A, benefit B, benefit C - **Learn more**: [Link](url)
After (Concise):
markdown### Tool Name Description of the tool that does X, Y, and Z to achieve A, B, and C. Use it when you need X by following steps 1-5. [Learn more](url)
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 11,273 | 15,474 | +37% | 1 | 1 | 0% | 2,206 | 3,139 | +42% | 0 | 0 | — |
case-02 | fail→pass | 9,273 | 8,406 | -9% | 1 | 1 | 0% | 1,526 | 3,320 | +118% | 0 | 0 | — |
case-03 | pass→pass | 5,299 | 4,535 | -14% | 1 | 1 | 0% | 1,017 | 2,465 | +142% | 0 | 0 | — |
case-04 | pass→pass | 8,149 | 4,331 | -47% | 1 | 1 | 0% | 1,341 | 2,519 | +88% | 0 | 0 | — |
case-05 | pass→fail | 10,748 | 12,154 | +13% | 1 | 1 | 0% | 1,953 | 3,884 | +99% | 0 | 0 | — |
case-06 | fail→pass | 7,096 | 5,928 | -16% | 1 | 1 | 0% | 1,530 | 2,929 | +91% | 0 | 0 | — |
case-07 | pass→pass | 9,714 | 5,716 | -41% | 1 | 1 | 0% | 821 | 2,775 | +238% | 0 | 0 | — |
case-08 | pass→pass | 16,372 | 17,405 | +6% | 1 | 1 | 0% | 2,710 | 4,218 | +56% | 0 | 0 | — |
case-09 | pass→pass | 11,554 | 11,078 | -4% | 1 | 1 | 0% | 2,057 | 3,462 | +68% | 0 | 0 | — |
case-10 | fail→pass | 9,250 | 5,885 | -36% | 1 | 1 | 0% | 1,359 | 2,838 | +109% | 0 | 0 | — |
case-11 | pass→pass | 6,965 | 3,274 | -53% | 1 | 1 | 0% | 1,127 | 2,229 | +98% | 0 | 0 | — |
case-12 | fail→pass | 8,466 | 3,271 | -61% | 1 | 1 | 0% | 1,393 | 2,348 | +69% | 0 | 0 | — |
case-13 | pass→pass | 8,275 | 8,852 | +7% | 1 | 1 | 0% | 1,320 | 3,378 | +156% | 0 | 0 | — |
case-14 | pass→pass | 3,942 | 3,216 | -18% | 1 | 1 | 0% | 655 | 2,346 | +258% | 0 | 0 | — |
case-24 | pass→pass | 10,779 | 7,932 | -26% | 1 | 1 | 0% | 1,889 | 3,091 | +64% | 0 | 0 | — |
case-15 | fail→pass | 9,991 | 4,482 | -55% | 1 | 1 | 0% | 2,248 | 2,504 | +11% | 0 | 0 | — |
case-16 | pass→pass | 19,531 | 14,722 | -25% | 1 | 1 | 0% | 3,250 | 3,956 | +22% | 0 | 0 | — |
case-17 | pass→pass | 4,777 | 4,271 | -11% | 1 | 1 | 0% | 898 | 2,456 | +173% | 0 | 0 | — |
case-18 | fail→pass | 7,523 | 6,926 | -8% | 1 | 1 | 0% | 1,329 | 2,892 | +118% | 0 | 0 | — |
case-25 | fail→pass | 17,633 | 14,228 | -19% | 1 | 1 | 0% | 2,849 | 3,876 | +36% | 0 | 0 | — |
case-19 | pass→pass | 9,402 | 18,807 | +100% | 1 | 1 | 0% | 1,798 | 3,983 | +122% | 0 | 0 | — |
case-20 | pass→pass | 16,507 | 10,654 | -35% | 1 | 1 | 0% | 3,072 | 3,777 | +23% | 0 | 0 | — |
case-21 | pass→pass | 11,386 | 8,692 | -24% | 1 | 1 | 0% | 1,968 | 3,404 | +73% | 0 | 0 | — |
case-22 | pass→pass | 6,273 | 4,628 | -26% | 1 | 1 | 0% | 798 | 2,419 | +203% | 0 | 0 | — |
case-23 | fail→fail | 4,632 | 5,092 | +10% | 1 | 1 | 0% | 603 | 2,538 | +321% | 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. 25 cases were attempted. The headline lift of +28 percentage points is the difference between those two pass rates over the 25 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.