| name | health-score |
| description | This skill should be used when the user asks to "check project health", "run health score", "health check", "プロジェクトの健康状態", "ヘルススコア", "健全性チェック", or on a schedule via cowork. Operates independently from review skills.
|
Health Score: Project Health Tracking
Track project health metrics over time. Independent from review execution.
Prerequisites:
- config.md exists with project information and tool sources
- workspace/ contains the project clone
Metrics Collection
1. Script-Based Metrics (Deterministic)
Run the health metrics script:
${CLAUDE_PLUGIN_ROOT}/scripts/health-metrics.sh <workspace-path>
This collects:
- File churn rates (last 30 days)
- Directory churn concentration
- Large file indicators
- Test-to-source ratio
- Fix commit frequency
2. External Tool Metrics
Check config.md for available tools. For each that exists, collect:
Sentry (if configured):
- Error rate: new errors in last 7 days vs previous 7 days
- Regression count: previously resolved errors that reappeared
- Use MCP or API if available
Datadog (if configured):
- Latency P95 trend: last 7 days vs previous 7 days
- Error rate trend
- Use MCP or API if available
SonarQube (if configured):
- Technical debt score change
- New security hotspots
- Use API if available
Bug Tracker (JIRA/GitHub Issues, as configured):
- New bugs filed in last 7 days
- Bug resolution rate
- Bug density by component/label
- Source identified during interview Phase 1
Skip any tool that isn't configured or accessible. Health score works with whatever is available.
3. Review Finding Accumulation
If previous review reports exist:
- Count Suggestion-level findings across recent reviews
- Track which areas accumulate repeated Suggestions
- Flag areas where Suggestion count is growing
Analysis (LLM)
After collecting all available metrics, analyze:
- Relative changes: Compare with previous health score (if exists in health/scores/)
- Trend detection: Is any metric consistently worsening?
- Hotspot identification: Which areas concentrate problems?
- Correlation: Does high churn correlate with high bug rate in the same area?
Output
Write to health/scores/YYYY-MM-DD.md:
# Health Score: [date]
## Summary
- Overall trend: improving / stable / degrading
- Areas of concern: [count]
## Metrics
### Code Churn
[Top churning files/dirs with change counts]
[Trend vs last score: ↑/↓/→]
### Bug Activity
[New bugs, resolution rate, density]
[Trend vs last score]
### Fix Ratio
[Fix commits as % of total]
[Trend vs last score]
### Test Coverage Indicator
[Test-to-source ratio]
[Trend vs last score]
### External Tools
[Sentry/Datadog/SonarQube metrics, if available]
### Accumulated Review Findings
[Areas with growing Suggestion counts]
## Areas of Concern
[Specific areas where multiple metrics are degrading]
## Recommendations
[Actionable suggestions: "Consider reviewing src/payments/ — churn rate is 3x average
and bug density is increasing"]
Report to user with key findings summary.