Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate comprehensive, accurate API documentation from source code. Use when creating or updating API documentation, generating OpenAPI specs, or when users mention API docs, endpoints, or documentation.
.claude/skills/luongnv89-doc-generator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -14% | 0% |
`markdown## GET /api/v1/users/:id ### 描述 简要说明这个端点做什么 ### 参数 | 名称 | 类型 | 必填 | 说明 | |------|------|------|------| | id | string | 是 | 用户 ID | ### 响应 **200 成功**
{ "id": "usr_123", "name": "John Doe", "email": "john@example.com", "created_at": "2025-01-15T10:30:00Z" }
**404 未找到**{ "error": "USER_NOT_FOUND", "message": "User does not exist" }
### 示例
**cURL**curl -X GET "https://api.example.com/api/v1/users/usr_123" \ -H "Authorization: Bearer YOUR_TOKEN"
**JavaScript**const user = await fetch('/api/v1/users/usr_123', { headers: { 'Authorization': 'Bearer token' } }).then(r => r.json());
**Python**response = requests.get( 'https://api.example.com/api/v1/users/usr_123', headers={'Authorization': 'Bearer token'} ) user = response.json()
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,182 | 9,522 | -33% | 1 | 1 | 0% | 3,332 | 2,606 | -22% | 0 | 0 | — |
case-17 | fail→pass | 11,305 | 6,698 | -41% | 1 | 1 | 0% | 2,589 | 1,888 | -27% | 0 | 0 | — |
case-02 | fail→pass | 11,713 | 6,884 | -41% | 1 | 1 | 0% | 2,804 | 1,849 | -34% | 0 | 0 | — |
case-03 | fail→fail | 13,160 | 7,248 | -45% | 1 | 1 | 0% | 3,142 | 1,958 | -38% | 0 | 0 | — |
case-04 | fail→pass | 13,721 | 6,698 | -51% | 1 | 1 | 0% | 3,168 | 2,006 | -37% | 0 | 0 | — |
case-05 | fail→fail | 11,070 | 6,334 | -43% | 1 | 1 | 0% | 2,869 | 1,879 | -35% | 0 | 0 | — |
case-06 | fail→pass | 15,428 | 6,097 | -60% | 1 | 1 | 0% | 3,317 | 1,736 | -48% | 0 | 0 | — |
case-07 | fail→fail | 13,049 | 6,871 | -47% | 1 | 1 | 0% | 2,617 | 1,916 | -27% | 0 | 0 | — |
case-08 | fail→fail | 13,758 | 8,483 | -38% | 1 | 1 | 0% | 2,921 | 2,277 | -22% | 0 | 0 | — |
case-09 | fail→pass | 12,635 | 9,133 | -28% | 1 | 1 | 0% | 2,940 | 2,541 | -14% | 0 | 0 | — |
case-10 | fail→fail | 14,793 | 7,341 | -50% | 1 | 1 | 0% | 3,116 | 2,067 | -34% | 0 | 0 | — |
case-11 | fail→pass | 13,475 | 7,210 | -46% | 1 | 1 | 0% | 3,006 | 2,054 | -32% | 0 | 0 | — |
case-12 | fail→pass | 13,047 | 7,922 | -39% | 1 | 1 | 0% | 2,875 | 2,185 | -24% | 0 | 0 | — |
case-13 | fail→pass | 12,988 | 7,652 | -41% | 1 | 1 | 0% | 2,830 | 2,041 | -28% | 0 | 0 | — |
case-14 | fail→pass | 11,387 | 6,802 | -40% | 1 | 1 | 0% | 2,956 | 1,913 | -35% | 0 | 0 | — |
case-15 | fail→pass | 10,164 | 3,702 | -64% | 1 | 1 | 0% | 2,224 | 1,275 | -43% | 0 | 0 | — |
case-16 | fail→pass | 10,431 | 8,252 | -21% | 1 | 1 | 0% | 2,720 | 2,200 | -19% | 0 | 0 | — |
case-18 | fail→fail | 14,821 | 8,494 | -43% | 1 | 1 | 0% | 3,353 | 2,283 | -32% | 0 | 0 | — |
case-19 | fail→fail | 14,409 | 6,751 | -53% | 1 | 1 | 0% | 3,229 | 2,170 | -33% | 0 | 0 | — |
case-20 | pass→fail | 19,803 | 15,006 | -24% | 1 | 1 | 0% | 3,990 | 3,657 | -8% | 0 | 0 | — |
case-21 | pass→pass | 17,786 | 14,545 | -18% | 1 | 1 | 0% | 3,955 | 3,871 | -2% | 0 | 0 | — |
case-22 | pass→pass | 13,888 | 10,965 | -21% | 1 | 1 | 0% | 3,383 | 3,078 | -9% | 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 +45 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 | 8/7/2026 | +41% |
Other measured skills in the registry, with their headline benchmark lift.