health-aggregate-skill

Health-Check Aggregation Phase

The third phase of the /assist:health-check workflow. Aggregates individual scores into a comprehensive health report.

Purpose

Combine individual component scores into an overall plugin health assessment with actionable recommendations.

Aggregation Process

1. Collect scores: Gather all component scores from analyze phase
2. Calculate averages: Weighted average based on component importance
3. Identify patterns: Find common issues across components
4. Generate recommendations: Prioritize improvements by impact

Weighting

Report Sections

Summary

• Overall health score and grade
• Total components analyzed
• Grade distribution
• Critical issues count

Component Details

• Individual scores and grades
• Per-component issues
• Per-component recommendations

Pattern Analysis

• Common issues found across multiple components
• Systemic problems
• Best practices violations

Recommendations

Prioritized list of improvements:

1. Critical (blocks deployment)
2. High (significantly impacts quality)
3. Medium (improves quality)
4. Low (nice to have)

Output Format

HEALTH_ANALYSIS_REPORT
======================

SUMMARY:
- Total Components: 11
- Overall Health Score: 85/100
- Overall Grade: B

GRADE_DISTRIBUTION:
- A (90-100): 4 components
- B (80-89): 5 components  
- C (70-79): 2 components
- D (60-69): 0 components
- F (< 60): 0 components

COMPONENT_REPORTS:
[Individual component scores...]

PATTERNS_DETECTED:
- 3 skills missing examples section
- 2 agents have overly broad tool lists

PRIORITIZED_RECOMMENDATIONS:

[HIGH] Add examples to skills
- Affects: router-skill, semantic-skill, execute-skill
- Impact: Improves Claude's understanding

[MEDIUM] Refine agent tool lists
- Affects: execute-agent, verify-agent
- Impact: Reduces potential for tool misuse

[LOW] Add more trigger phrases
- Affects: All skills
- Impact: Improves skill discovery

Transition Evidence

To proceed to schema-check phase, provide:

• overall_score: Weighted average health score
• overall_grade: Letter grade
• critical_count: Number of critical issues (must be 0)
• recommendations: Prioritized recommendation list