← ClaudeAtlas

monitor-platform-healthlisted

SOP for measuring API/website health via machincy-grokbot - uptime events, security events, app errors, and write integrity. Read and list oriented; does not restart services or block IPs.
Chuck3PointZero/OAA · ★ 1 · API & Backend · score 78
Install: claude install-skill Chuck3PointZero/OAA
# Skill: Monitor Platform Health Produces a **platform-health packet** for the ops digest. Complements (does not replace) webservice-monitor's api-health-check and api-smoke-test; this skill aggregates list/report surfaces the Operations Manager needs for one daily view. ## Workflow 1. **Uptime.** Call `machincy-grokbot__list_uptime_events` (e.g. since last 24h). Note any `health_check_failure` or `smoke_test_failure` rows and whether they were `notified`. 2. **Security.** Call `machincy-grokbot__list_security_events` (or equivalent list) for the window. Count confirmed vs noise if the payload distinguishes them. Do not re-scan raw access logs unless investigating a specific incident already escalated - deep access-log audit belongs to security-auditor. 3. **App errors.** Call list app errors for the window. Compute `error_count` and whether it looks like a spike vs routine (use baseline from memory or instructions when present). 4. **Write integrity.** Call get write integrity report. Flag any unattributed or mismatched counts the API returns. 5. **Packet fields:** - `uptime_events_count` / `uptime_unnotified_count` - `security_events_count` - `app_error_count` / `app_error_spike` (boolean or note) - `write_integrity_ok` (boolean or short status) - `anomalies` - named issues for evaluate-and-flag-issues 6. **Skill-level failure.** If a list call fails (non-2xx / timeout / unparseable), record which capability failed in the packet and treat that capabil