Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Technical documentation expert for creating clear, comprehensive documentation. Use when user asks to write docs, create README, or document code.
.claude/skills/documentation-engineer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 100% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -41% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -16% | 0% |
Expert in creating clear, comprehensive, and maintainable technical documentation.
Activates when you:
Every project should have a README with:
markdown# Project Name Brief description (what it does, why it exists) ## Quick Start Installation and usage in 3 steps or less. ## Installation Detailed installation instructions. ## Usage Examples of common usage patterns. ## Configuration Environment variables and configuration options. ## Development How to run tests, build, and develop locally. ## Contributing Guidelines for contributors. ## License License information.
For each endpoint/function:
Comment why, not what:
typescript// Bad: Sets the count to zero count = 0; // Good: Reset count for new measurement cycle count = 0; // Bad: Check if user is admin if (user.role === 'admin') { // Good: Only admins can bypass approval workflow if (user.role === 'admin') {
Generate documentation structure:
bashpython3 scripts/generate_docs.py --name <service-name> --output docs/README.md
Validate documentation:
bashpython3 scripts/validate_docs.py --input docs/README.md
references/readme-template.md - README templatereferences/api-template.md - API documentation templatereferences/style-guide.md - Documentation style guide| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 21,427 | 19,887 | -7% | 1 | 1 | 0% | 3,048 | 3,550 | +16% | 0 | 0 | — |
case-02 | fail→fail | 29,614 | 28,905 | -2% | 1 | 1 | 0% | 3,804 | 5,531 | +45% | 0 | 0 | — |
case-03 | fail→fail | 33,040 | 21,364 | -35% | 1 | 1 | 0% | 3,930 | 3,443 | -12% | 0 | 0 | — |
case-04 | fail→fail | 23,998 | 22,748 | -5% | 1 | 1 | 0% | 3,340 | 4,231 | +27% | 0 | 0 | — |
case-05 | pass→pass | 22,949 | 27,346 | +19% | 1 | 1 | 0% | 2,746 | 3,986 | +45% | 0 | 0 | — |
case-06 | pass→pass | 21,306 | 23,758 | +12% | 1 | 1 | 0% | 3,546 | 4,426 | +25% | 0 | 0 | — |
case-07 | fail→pass | 11,085 | 10,596 | -4% | 1 | 1 | 0% | 808 | 1,614 | +100% | 0 | 0 | — |
case-08 | pass→pass | 12,928 | 16,194 | +25% | 1 | 1 | 0% | 1,193 | 2,517 | +111% | 0 | 0 | — |
case-09 | fail→pass | 16,414 | 2,578 | -84% | 1 | 1 | 0% | 1,929 | 1,147 | -41% | 0 | 0 | — |
case-10 | fail→pass | 8,072 | 6,850 | -15% | 1 | 1 | 0% | 1,300 | 998 | -23% | 0 | 0 | — |
case-11 | fail→fail | 12,909 | 18,705 | +45% | 1 | 1 | 0% | 2,116 | 2,699 | +28% | 0 | 0 | — |
case-12 | pass→pass | 17,440 | 11,151 | -36% | 1 | 1 | 0% | 2,351 | 2,855 | +21% | 0 | 0 | — |
case-13 | fail→fail | 11,422 | 12,264 | +7% | 1 | 1 | 0% | 2,251 | 2,912 | +29% | 0 | 0 | — |
case-14 | pass→pass | 9,354 | 8,660 | -7% | 1 | 1 | 0% | 628 | 1,240 | +97% | 0 | 0 | — |
case-15 | fail→pass | 17,998 | 1,840 | -90% | 1 | 1 | 0% | 1,616 | 977 | -40% | 0 | 0 | — |
case-16 | fail→pass | 7,643 | 1,498 | -80% | 1 | 1 | 0% | 1,171 | 978 | -16% | 0 | 0 | — |
case-17 | fail→pass | 9,680 | 2,008 | -79% | 1 | 1 | 0% | 909 | 977 | +7% | 0 | 0 | — |
case-18 | pass→pass | 39,942 | 24,194 | -39% | 1 | 1 | 0% | 5,711 | 3,553 | -38% | 0 | 0 | — |
case-19 | pass→pass | 14,324 | 19,670 | +37% | 1 | 1 | 0% | 2,705 | 3,044 | +13% | 0 | 0 | — |
case-20 | fail→fail | 15,415 | 11,941 | -23% | 1 | 1 | 0% | 3,042 | 3,269 | +7% | 0 | 0 | — |
case-21 | pass→pass | 12,307 | 9,959 | -19% | 1 | 1 | 0% | 1,261 | 1,527 | +21% | 0 | 0 | — |
case-22 | fail→pass | 11,655 | 10,729 | -8% | 1 | 1 | 0% | 1,720 | 1,645 | -4% | 0 | 0 | — |
case-23 | pass→pass | 10,296 | 9,107 | -12% | 1 | 1 | 0% | 1,388 | 1,405 | +1% | 0 | 0 | — |
case-24 | pass→pass | 14,626 | 7,198 | -51% | 1 | 1 | 0% | 1,367 | 1,717 | +26% | 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. 24 cases were attempted. The headline lift of +29 percentage points is the difference between those two pass rates over the 24 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 7/26/2026 | +39% |
Other measured skills in the registry, with their headline benchmark lift.