PrizmKit Monitoring Setup

Generate comprehensive monitoring, alerting, and log collection configurations for your project's observability stack.

Commands

prizmkit.monitoring

Generate monitoring and observability configurations.

STEPS:

1. Read .prizm-docs/root.prizm for tech stack and architecture (services, databases, external dependencies)
2. Ask user: monitoring stack
   • Prometheus + Grafana
   • ELK (Elasticsearch, Logstash, Kibana)
   • CloudWatch (AWS)
   • Datadog
   • Other (specify)
3. Identify key metrics to monitor:
   • Application (RED metrics):
     • Request Rate: requests per second by endpoint
     • Error rate: 4xx/5xx responses, exception counts
     • Duration: latency percentiles (p50, p95, p99)
   • System:
     • CPU utilization and saturation
     • Memory usage and swap
     • Disk I/O and space
     • Network throughput and errors
   • Business:
     • Feature-specific metrics derived from spec (signups, transactions, etc.)
     • SLA/SLO compliance indicators
   • Database:
     • Connection pool utilization
     • Query latency by type (read/write)
     • Replication lag
     • Slow query count
4. Generate monitoring configs:
   • Metrics collection: Scrape configs, exporters, or agent configurations
   • Alert rules with severity levels:
     • Critical: immediate page (service down, data loss risk)
     • Warning: investigate soon (degraded performance, approaching limits)
     • Info: awareness (deployment events, scaling events)
   • Dashboard definition: JSON/YAML for Grafana or equivalent
     • Overview dashboard: service health at a glance
     • Detail dashboard: per-service deep dive
   • Log format and collection config:
     • Structured log format recommendation
     • Log shipping configuration
     • Log retention policy
5. Generate health check endpoint code if not existing:
   • Liveness probe: process is running
   • Readiness probe: dependencies are available
   • Startup probe: initialization complete
6. Write configs to standard locations:
   • Prometheus: monitoring/prometheus.yml, monitoring/alerts.yml
   • Grafana: monitoring/dashboards/
   • Application: health check endpoint in source code

Path References

All internal asset paths MUST use ${SKILL_DIR} placeholder for cross-IDE compatibility.

Output

• Monitoring configuration files in monitoring/ directory
• Alert rule definitions
• Dashboard JSON/YAML files
• Health check endpoint code (if generated)