Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Research specialization domains, compile references, analyze best practices, and gather comprehensive knowledge for new specialization creation.
.claude/skills/a5c-ai-specialization-researcher/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 7 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 23% | 0% |
You are specialization-researcher - a specialized skill for researching and gathering comprehensive knowledge about specialization domains within the Babysitter SDK framework.
This skill enables systematic research of specialization domains including:
Research the specialization domain thoroughly:
Gather and organize reference materials:
Identify and document best practices:
Identify roles and responsibilities:
javascript{ task: 'Research the data engineering domain', domain: 'data-engineering', scope: ['ETL', 'data pipelines', 'analytics'], outputFormat: 'README and references' }
javascript{ task: 'Compile references for machine learning', domain: 'machine-learning', referenceTypes: ['papers', 'tutorials', 'tools'], maxReferences: 50 }
json{ "domain": "specialization-name", "overview": "Comprehensive domain overview", "roles": [ { "name": "Role Name", "responsibilities": ["resp1", "resp2"], "skills": ["skill1", "skill2"] } ], "references": [ { "title": "Reference Title", "url": "https://...", "category": "documentation", "description": "Brief description" } ], "bestPractices": ["practice1", "practice2"], "artifacts": ["README.md", "references.md"] }
This skill integrates with:
specialization-creation.js - Phase 1 researchphase1-research-readme.js - README generationdomain-creation.js - Domain research| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 28,916 | 22,562 | -22% | 1 | 1 | 0% | 4,499 | 4,062 | -10% | 0 | 0 | — |
case-02 | fail→pass | 20,511 | 19,035 | -7% | 1 | 1 | 0% | 3,607 | 3,340 | -7% | 0 | 0 | — |
case-03 | fail→pass | 26,161 | 18,516 | -29% | 1 | 1 | 0% | 4,710 | 3,632 | -23% | 0 | 0 | — |
case-04 | pass→pass | 26,997 | 28,422 | +5% | 1 | 1 | 0% | 6,035 | 5,389 | -11% | 0 | 0 | — |
case-05 | pass→pass | 21,088 | 24,879 | +18% | 1 | 1 | 0% | 4,123 | 6,054 | +47% | 0 | 0 | — |
case-06 | fail→pass | 20,388 | 21,221 | +4% | 1 | 1 | 0% | 4,489 | 5,437 | +21% | 0 | 0 | — |
case-07 | fail→pass | 16,233 | 18,254 | +12% | 1 | 1 | 0% | 2,875 | 3,547 | +23% | 0 | 0 | — |
case-08 | fail→pass | 17,220 | 11,904 | -31% | 1 | 1 | 0% | 2,886 | 2,811 | -3% | 0 | 0 | — |
case-09 | fail→pass | 41,073 | 14,661 | -64% | 1 | 1 | 0% | 8,061 | 3,392 | -58% | 0 | 0 | — |
case-10 | fail→pass | 21,616 | 15,037 | -30% | 1 | 1 | 0% | 3,303 | 3,325 | +1% | 0 | 0 | — |
case-11 | fail→pass | 40,618 | 11,844 | -71% | 1 | 1 | 0% | 6,126 | 2,957 | -52% | 0 | 0 | — |
case-12 | fail→pass | 26,288 | 16,375 | -38% | 1 | 1 | 0% | 4,621 | 3,752 | -19% | 0 | 0 | — |
case-13 | fail→pass | 28,419 | 14,464 | -49% | 1 | 1 | 0% | 5,613 | 3,362 | -40% | 0 | 0 | — |
case-14 | fail→pass | 30,364 | 14,384 | -53% | 1 | 1 | 0% | 5,724 | 3,527 | -38% | 0 | 0 | — |
case-15 | fail→pass | 20,207 | 18,769 | -7% | 1 | 1 | 0% | 3,633 | 3,825 | +5% | 0 | 0 | — |
case-16 | fail→pass | 22,752 | 16,335 | -28% | 1 | 1 | 0% | 4,415 | 3,494 | -21% | 0 | 0 | — |
case-17 | fail→pass | 32,836 | 19,818 | -40% | 1 | 1 | 0% | 6,198 | 4,353 | -30% | 0 | 0 | — |
case-18 | pass→pass | 43,435 | 15,591 | -64% | 1 | 1 | 0% | 7,302 | 3,194 | -56% | 0 | 0 | — |
case-19 | pass→pass | 42,424 | 12,699 | -70% | 1 | 1 | 0% | 8,224 | 3,191 | -61% | 0 | 0 | — |
case-20 | fail→pass | 14,948 | 14,498 | -3% | 1 | 1 | 0% | 2,535 | 3,196 | +26% | 0 | 0 | — |
case-21 | fail→pass | 15,590 | 21,272 | +36% | 1 | 1 | 0% | 2,744 | 3,284 | +20% | 0 | 0 | — |
case-22 | fail→pass | 15,372 | 13,882 | -10% | 1 | 1 | 0% | 2,760 | 2,785 | +1% | 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 +82 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.