Health scoring engine with 4 dimensions: Gateway stability (40%), config
completeness (25%), skill consistency (20%), and Cron success rate (15%).
Grades A(80-100) B(60-79) C(40-59) D(0-39) with detailed deduction items
and fix suggestions per dimension.
API endpoints:
- GET /health/scores — list all profiles with grade filter + sort
- GET /health/scores/{profile} — detail with deductions + suggestions
- POST /health/compare — multi-profile config/skill comparison
- POST /health/calculate — manual recalculation (owner only)
Frontend HealthPage: grade distribution cards, score table with expandable
detail rows, batch compare view with diff highlighting, grade/select/sort
filters, and owner-only recalculation.
Data sources use placeholders (TODO: hermes CLI / file reads) but the full
scoring logic is implemented end-to-end.
Co-Authored-By: Claude <noreply@anthropic.com>