Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create new agent skills with proper structure, progressive disclosure, and bundled resources. Use when user wants to create, write, or build a new skill.
.claude/skills/mkurman-write-a-skill/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -5% | 0% |
Minimal working example]
Step-by-step processes with checklists for complex tasks]
Link to separate files: See REFERENCE.md]
## Description Requirements
The description is **the only thing your agent sees** when deciding which skill to load. It's surfaced in the system prompt alongside all other installed skills. Your agent reads these descriptions and picks the relevant skill based on the user's request.
**Goal**: Give your agent just enough info to know:
1. What capability this skill provides
2. When/why to trigger it (specific keywords, contexts, file types)
**Format**:
- Max 1024 chars
- Write in third person
- First sentence: what it does
- Second sentence: "Use when [specific triggers]"
**Good example**:
Extract text and tables from PDF files, fill forms, merge documents. Use when working with PDF files or when user mentions PDFs, forms, or document extraction.
**Bad example**:
Helps with documents.
The bad example gives your agent no way to distinguish this from other document skills.
## When to Add Scripts
Add utility scripts when:
- Operation is deterministic (validation, formatting)
- Same code would be generated repeatedly
- Errors need explicit handling
Scripts save tokens and improve reliability vs generated code.
## When to Split Files
Split into separate files when:
- SKILL.md exceeds 100 lines
- Content has distinct domains (finance vs sales schemas)
- Advanced features are rarely needed
## Review Checklist
After drafting, verify:
- [ ] Description includes triggers ("Use when...")
- [ ] SKILL.md under 100 lines
- [ ] No time-sensitive info
- [ ] Consistent terminology
- [ ] Concrete examples included
- [ ] References one level deep| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 12,546 | 3,969 | -68% | 1 | 1 | 0% | 1,969 | 1,018 | -48% | 0 | 0 | — |
case-02 | fail→pass | 8,926 | 7,761 | -13% | 1 | 1 | 0% | 1,159 | 1,682 | +45% | 0 | 0 | — |
case-03 | fail→pass | 11,369 | 5,784 | -49% | 1 | 1 | 0% | 1,693 | 1,313 | -22% | 0 | 0 | — |
case-04 | pass→pass | 9,122 | 5,565 | -39% | 1 | 1 | 0% | 1,353 | 1,147 | -15% | 0 | 0 | — |
case-05 | pass→pass | 13,202 | 5,569 | -58% | 1 | 1 | 0% | 2,111 | 1,284 | -39% | 0 | 0 | — |
case-06 | pass→pass | 12,056 | 7,958 | -34% | 1 | 1 | 0% | 1,761 | 1,478 | -16% | 0 | 0 | — |
case-07 | pass→pass | 15,707 | 10,795 | -31% | 1 | 1 | 0% | 2,471 | 2,325 | -6% | 0 | 0 | — |
case-08 | pass→pass | 11,577 | 7,885 | -32% | 1 | 1 | 0% | 1,750 | 1,673 | -4% | 0 | 0 | — |
case-09 | pass→pass | 13,368 | 4,866 | -64% | 1 | 1 | 0% | 1,920 | 1,129 | -41% | 0 | 0 | — |
case-10 | pass→pass | 11,631 | 4,724 | -59% | 1 | 1 | 0% | 1,643 | 1,080 | -34% | 0 | 0 | — |
case-11 | pass→pass | 17,671 | 5,981 | -66% | 1 | 1 | 0% | 1,586 | 1,351 | -15% | 0 | 0 | — |
case-12 | fail→pass | 8,863 | 4,982 | -44% | 1 | 1 | 0% | 1,446 | 1,227 | -15% | 0 | 0 | — |
case-13 | pass→pass | 13,588 | 9,118 | -33% | 1 | 1 | 0% | 2,207 | 1,837 | -17% | 0 | 0 | — |
case-14 | pass→pass | 11,995 | 2,565 | -79% | 1 | 1 | 0% | 1,678 | 678 | -60% | 0 | 0 | — |
case-15 | pass→pass | 12,591 | 6,904 | -45% | 1 | 1 | 0% | 1,843 | 1,475 | -20% | 0 | 0 | — |
case-16 | pass→pass | 11,164 | 7,832 | -30% | 1 | 1 | 0% | 1,681 | 1,595 | -5% | 0 | 0 | — |
case-17 | pass→pass | 9,671 | 4,524 | -53% | 1 | 1 | 0% | 1,330 | 995 | -25% | 0 | 0 | — |
case-18 | fail→pass | 10,752 | 7,116 | -34% | 1 | 1 | 0% | 1,606 | 1,519 | -5% | 0 | 0 | — |
case-19 | pass→pass | 7,135 | 4,146 | -42% | 1 | 1 | 0% | 1,171 | 1,050 | -10% | 0 | 0 | — |
case-20 | pass→pass | 13,343 | 7,548 | -43% | 1 | 1 | 0% | 1,579 | 1,520 | -4% | 0 | 0 | — |
case-21 | fail→pass | 11,115 | 4,956 | -55% | 1 | 1 | 0% | 1,727 | 1,156 | -33% | 0 | 0 | — |
case-22 | pass→pass | 10,809 | 10,150 | -6% | 1 | 1 | 0% | 1,667 | 1,922 | +15% | 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 +27 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.