alertslisted

# Alerts -- Lightweight Health Signal **Ecosystem role:** Consumer of `health-ecosystem-audit` ecosystem (D#17). `/health-ecosystem-audit` OWNS the ecosystem. `/ecosystem-health` and `/alerts` CONSUME it. `/audit-health` audits the audit system, not health. ## Critical Rules 1. **MUST** run the checker script before presenting any data 2. **MUST** present every alert to the user (no silent filtering) 3. **MUST** recommend an action per alert based on severity + trend 4. **MUST** log every decision to session JSONL immediately (per-alert, not batched) 5. **MUST NOT** present a health score without running checkers first 6. **MUST** surface previous learnings from alerts-history.jsonl at warm-up ## When to Use - User explicitly invokes `/alerts` - User asks "check alerts", "what needs attention", "system health", "show warnings", "pending issues" - Called from `session-begin` with `--limited` mode ## When NOT to Use | Need | Use Instead | Why | | ----------------------- | ------------------- | ---------------------------------- | | Domain-specific audit | `/audit-code`, etc. | Deeper analysis per domain | | Audit system meta-check | `/audit-health` | Checks audit infra, not health | | Full health dashboard | `/ecosystem-health` | 8 weighted categories, deep triage | | Skill validation | `/skill-audit` | Quality audit for individual skill | ## Routing Decision Tree - "Just sho