Install any skill in seconds. Free to start, no credit card required.
Get Started Free →You are a documentation expert specializing in creating comprehensive, maintainable documentation from code. Generate API docs, architecture diagrams, user guides, and technical references using AI-powered analysis and industry best practices.
.claude/skills/documentation-generation-doc-generate/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 19 |
| gemini-3.1-pro-preview | 100% | 1 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-06 | ✓→✗ | ▼ Worse | -19% | 0% |
Primary editorial path for this compatibility group. The full instructions and support files remain local so existing installations continue to work offline. This is one shared procedure, not an additional capability. Preserve the callable ID when an existing manifest or client configuration uses it. Modified in AAS on 2026-09-05; original metadata and license notices are retained.
You are a documentation expert specializing in creating comprehensive, maintainable documentation from code. Generate API docs, architecture diagrams, user guides, and technical references using AI-powered analysis and industry best practices.
The user needs automated documentation generation that extracts information from code, creates clear explanations, and maintains consistency across documentation types. Focus on creating living documentation that stays synchronized with code.
$ARGUMENTS
resources/implementation-playbook.md.For an endpoint changed to return a cursor, inspect its implementation and test fixtures, update the response example and pagination explanation, then run the existing schema/doc build and the documented read-only call against a test fixture. Record which commands actually ran. The bundled playbook contains incomplete integration sketches, not an installed generator or tested project template.
resources/implementation-playbook.md for detailed examples and templates.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 35,105 | 28,767 | -18% | 1 | 1 | 0% | 3,617 | 4,321 | +19% | 0 | 0 | — |
case-02 | fail→fail | 38,524 | 10,594 | -73% | 1 | 1 | 0% | 5,678 | 1,229 | -78% | 0 | 0 | — |
case-03 | pass→pass | 13,606 | 11,232 | -17% | 1 | 1 | 0% | 1,578 | 1,512 | -4% | 0 | 0 | — |
case-04 | fail→fail | 192,392 | 12,093 | -94% | 1 | 1 | 0% | 5,578 | 2,708 | -51% | 0 | 0 | — |
case-05 | fail→pass | 19,232 | 18,323 | -5% | 1 | 1 | 0% | 2,442 | 2,638 | +8% | 0 | 0 | — |
case-06 | pass→fail | 33,191 | 20,119 | -39% | 1 | 1 | 0% | 4,318 | 3,485 | -19% | 0 | 0 | — |
case-07 | pass→pass | 21,122 | 18,802 | -11% | 1 | 1 | 0% | 3,705 | 3,330 | -10% | 0 | 0 | — |
case-08 | pass→pass | 25,951 | 12,400 | -52% | 1 | 1 | 0% | 2,063 | 2,727 | +32% | 0 | 0 | — |
case-09 | fail→fail | 8,318 | 18,235 | +119% | 1 | 1 | 0% | 1,233 | 2,701 | +119% | 0 | 0 | — |
case-10 | pass→pass | 23,827 | 18,624 | -22% | 1 | 1 | 0% | 3,611 | 2,977 | -18% | 0 | 0 | — |
case-11 | fail→pass | 48,399 | 7,455 | -85% | 1 | 1 | 0% | 2,453 | 807 | -67% | 0 | 0 | — |
case-12 | pass→pass | 27,031 | 45,869 | +70% | 1 | 1 | 0% | 3,589 | 3,387 | -6% | 0 | 0 | — |
case-13 | fail→pass | 15,982 | 26,925 | +68% | 1 | 1 | 0% | 1,581 | 2,093 | +32% | 0 | 0 | — |
case-14 | pass→pass | 14,960 | 29,845 | +99% | 1 | 1 | 0% | 2,128 | 3,104 | +46% | 0 | 0 | — |
case-15 | pass→pass | 29,787 | 16,767 | -44% | 1 | 1 | 0% | 3,920 | 3,292 | -16% | 0 | 0 | — |
case-16 | pass→pass | 42,075 | 34,721 | -17% | 1 | 1 | 0% | 6,182 | 4,032 | -35% | 0 | 0 | — |
case-17 | pass→pass | 22,304 | 21,710 | -3% | 1 | 1 | 0% | 3,589 | 3,270 | -9% | 0 | 0 | — |
case-18 | pass→pass | 26,313 | 21,477 | -18% | 1 | 1 | 0% | 2,983 | 2,852 | -4% | 0 | 0 | — |
case-19 | fail→pass | 57,032 | 27,999 | -51% | 1 | 1 | 0% | 2,642 | 3,947 | +49% | 0 | 0 | — |
case-20 | pass→pass | 36,829 | 23,787 | -35% | 1 | 1 | 0% | 5,094 | 3,497 | -31% | 0 | 0 | — |
case-21 | pass→pass | 20,486 | 27,339 | +33% | 1 | 1 | 0% | 2,234 | 3,967 | +78% | 0 | 0 | — |
case-22 | fail→fail | 25,301 | 13,292 | -47% | 1 | 1 | 0% | 3,359 | 1,694 | -50% | 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 +14 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 7/28/2026 | +11% |
Other measured skills in the registry, with their headline benchmark lift.