Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Validate specialization completeness across all 7 phases, score each phase, identify gaps, and generate validation reports.
.claude/skills/a5c-ai-specialization-validator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 1488% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 70% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 24% | 0% |
You are specialization-validator - a specialized skill for validating Babysitter SDK specializations across all 7 phases of the creation workflow.
This skill validates specialization completeness including:
Validate README and references:
json{ "checks": [ "README.md exists", "README has Overview section", "README has Roles section", "README has Directory Structure", "references.md exists", "references.md has categorized links" ], "score": 90, "issues": ["Missing best practices section"] }
Validate processes backlog:
json{ "checks": [ "processes-backlog.md exists", "Has TODO format items", "Has process descriptions", "Processes are categorized" ], "processCount": 15, "score": 100, "issues": [] }
Validate process JS files:
json{ "checks": [ "JS files exist for backlog items", "Files have JSDoc metadata", "Files import defineTask", "Files export process function", "Tasks have proper structure" ], "processCount": 15, "implementedCount": 12, "score": 80, "issues": ["3 processes not implemented"] }
Validate skills/agents backlog:
json{ "checks": [ "skills-agents-backlog.md exists", "Skills section with SK-XX-NNN format", "Agents section with AG-XX-NNN format", "Process-to-Skill/Agent mapping table" ], "skillCount": 10, "agentCount": 5, "score": 100, "issues": [] }
Validate references file:
json{ "checks": [ "skills-agents-references.md exists", "Has external references", "Has GitHub links", "Has MCP server references" ], "referenceCount": 20, "score": 85, "issues": ["Missing MCP server section"] }
Validate skill and agent files:
json{ "checks": [ "skills/ directory exists", "agents/ directory exists", "SKILL.md files have valid frontmatter", "AGENT.md files have valid frontmatter" ], "skillCount": 10, "agentCount": 5, "createdSkills": 8, "createdAgents": 4, "score": 75, "issues": ["2 skills missing", "1 agent missing"] }
Validate integration:
json{ "checks": [ "Process files reference skills", "Process files reference agents", "References match backlog mapping" ], "totalTasks": 50, "integratedTasks": 45, "score": 90, "issues": ["5 tasks missing skill/agent references"] }
Each phase is scored 0-100 based on:
Overall score uses weighted average:
json{ "valid": true, "overallScore": 85, "phases": { "phase1": { "score": 90, "complete": true, "issues": [] }, "phase2": { "score": 100, "complete": true, "issues": [] }, "phase3": { "score": 80, "complete": false, "issues": ["3 missing"] }, "phase4": { "score": 100, "complete": true, "issues": [] }, "phase5": { "score": 85, "complete": true, "issues": [] }, "phase6": { "score": 75, "complete": false, "issues": ["3 missing"] }, "phase7": { "score": 90, "complete": true, "issues": [] } }, "gaps": ["phase3: 3 processes", "phase6: 2 skills, 1 agent"], "recommendations": ["Implement remaining processes", "Create missing skills"] }
This skill integrates with:
specialization-validator.js - Primary validation processbacklog-gap-analyzer.js - Gap analysisspecialization-creation.js - Post-creation validation| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 12,671 | 13,833 | +9% | 1 | 1 | 0% | 2,303 | 3,231 | +40% | 0 | 0 | — |
case-02 | fail→pass | 6,733 | 20,986 | +212% | 1 | 1 | 0% | 311 | 4,940 | +1488% | 0 | 0 | — |
case-03 | fail→fail | 9,683 | 8,293 | -14% | 1 | 1 | 0% | 1,800 | 3,181 | +77% | 0 | 0 | — |
case-04 | fail→fail | 5,923 | 6,194 | +5% | 1 | 1 | 0% | 916 | 2,400 | +162% | 0 | 0 | — |
case-05 | pass→fail | 4,978 | 22,043 | +343% | 1 | 1 | 0% | 778 | 4,914 | +532% | 0 | 0 | — |
case-06 | fail→pass | 12,577 | 12,252 | -3% | 1 | 1 | 0% | 1,880 | 3,197 | +70% | 0 | 0 | — |
case-07 | pass→pass | 9,726 | 9,102 | -6% | 1 | 1 | 0% | 1,605 | 2,764 | +72% | 0 | 0 | — |
case-08 | pass→pass | 13,599 | 8,037 | -41% | 1 | 1 | 0% | 1,831 | 2,581 | +41% | 0 | 0 | — |
case-09 | pass→pass | 16,664 | 14,964 | -10% | 1 | 1 | 0% | 2,429 | 3,735 | +54% | 0 | 0 | — |
case-10 | pass→pass | 10,914 | 5,080 | -53% | 1 | 1 | 0% | 1,491 | 2,492 | +67% | 0 | 0 | — |
case-11 | fail→pass | 12,298 | 8,584 | -30% | 1 | 1 | 0% | 1,618 | 2,782 | +72% | 0 | 0 | — |
case-12 | pass→pass | 6,981 | 6,186 | -11% | 1 | 1 | 0% | 957 | 2,689 | +181% | 0 | 0 | — |
case-13 | pass→pass | 10,067 | 6,851 | -32% | 1 | 1 | 0% | 1,395 | 2,413 | +73% | 0 | 0 | — |
case-14 | fail→pass | 12,191 | 6,387 | -48% | 1 | 1 | 0% | 2,260 | 2,805 | +24% | 0 | 0 | — |
case-15 | fail→pass | 21,476 | 2,085 | -90% | 1 | 1 | 0% | 2,904 | 1,754 | -40% | 0 | 0 | — |
case-16 | fail→pass | 8,576 | 1,889 | -78% | 1 | 1 | 0% | 1,156 | 1,706 | +48% | 0 | 0 | — |
case-17 | fail→pass | 5,210 | 3,323 | -36% | 1 | 1 | 0% | 589 | 1,852 | +214% | 0 | 0 | — |
case-18 | fail→pass | 9,773 | 3,915 | -60% | 1 | 1 | 0% | 1,659 | 2,149 | +30% | 0 | 0 | — |
case-19 | fail→pass | 7,272 | 3,560 | -51% | 1 | 1 | 0% | 1,165 | 2,122 | +82% | 0 | 0 | — |
case-20 | fail→pass | 13,902 | 2,647 | -81% | 1 | 1 | 0% | 2,284 | 1,773 | -22% | 0 | 0 | — |
case-21 | pass→pass | 10,688 | 2,621 | -75% | 1 | 1 | 0% | 1,819 | 1,850 | +2% | 0 | 0 | — |
case-22 | pass→pass | 10,909 | 5,829 | -47% | 1 | 1 | 0% | 1,623 | 2,239 | +38% | 0 | 0 | — |
case-23 | fail→pass | 13,965 | 5,514 | -61% | 1 | 1 | 0% | 2,042 | 2,128 | +4% | 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. 23 cases were attempted, and 22 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +48 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.
Other measured skills in the registry, with their headline benchmark lift.