sla-monitor

# SLA Monitor Skill ## Purpose Help teams set up production-grade monitoring for AI agents and automated services. Covers uptime tracking, response time SLAs, error budgets, and incident escalation. ## When to Use - Deploying AI agents to production - Setting up monitoring for client-facing automation - Creating SLA documentation for service agreements - Building incident response procedures ## Monitoring Stack Options ### Option 1: UptimeRobot (Free tier available) - 50 monitors free, 5-minute intervals - HTTP, keyword, ping, port monitors - Email + Slack + webhook alerts ### Option 2: Better Stack (Formerly Uptime.com) - Status pages included - Incident management built-in - Free tier: 10 monitors ### Option 3: Self-Hosted (Uptime Kuma) ```bash docker run -d --restart=always -p 3001:3001 -v uptime-kuma:/app/data --name uptime-kuma louislam/uptime-kuma:1 ``` ## SLA Tiers ### Tier 1: Standard ($1,500/mo) - 99.5% uptime guarantee (43.8h downtime/year) - Response within 4 hours (business hours) - Monthly performance report ### Tier 2: Professional ($3,000/mo) - 99.9% uptime guarantee (8.76h downtime/year) - Response within 1 hour (business hours) - Weekly performance reports - Quarterly optimization reviews ### Tier 3: Enterprise ($5,000+/mo) - 99.95% uptime (4.38h downtime/year) - Response within 15 minutes (24/7) - Real-time dashboard access - Dedicated support channel ## Alert Configuration Template ```yaml monitors: - name: "Agent Health Check" type: http url: "https://your-agent-endpoint/health" interval: 300 # 5 minutes alerts: - type: email threshold: 1 # alert after 1 failure - type: slack webhook: "${SLACK_WEBHOOK}" threshold: 2 # alert after 2 consecutive failures - type: sms threshold: 3 # escalate after 3 failures - name: "API Response Time" type: http url: "https://your-agent-endpoint/api" interval: 60 expected_response_time: 2000 # ms alerts: - type: slack condition: "response_time > 5000" error_budget: monthly_target: 99.9 burn_rate_alert: 2.0 # Alert if burning 2x normal rate ``` ## Incident Response Playbook ### Severity 1 — Total Outage 1. Acknowledge within 5 minutes 2. Status page update within 10 minutes 3. Root cause identification within 30 minutes 4. Resolution or workaround within 2 hours 5. Post-mortem within 24 hours ### Severity 2 — Degraded Performance 1. Acknowledge within 15 minutes 2. Investigation within 30 minutes 3. Resolution within 4 hours 4. Summary report within 48 hours ### Severity 3 — Minor Issue 1. Acknowledge within 1 hour 2. Resolution within 24 hours 3. Logged for next review cycle ## Error Budget Calculator ``` Monthly minutes: 43,200 (30 days) 99.9% SLA = 43.2 minutes downtime allowed 99.5% SLA = 216 minutes downtime allowed 99.0% SLA = 432 minutes downtime allowed Burn rate = (actual downtime / budget) × 100 If burn rate > 50% with 2+ weeks remaining → review needed If burn rate > 80% → freeze deployments ``` ## Status Page Template Provide clients with a public status page showing: - Current system status (operational / degraded / outage) - Component-level status (Agent A, Agent B, API, Dashboard) - Uptime percentage (30-day rolling) - Incident history with resolution notes - Scheduled maintenance windows ## Next Steps Need managed AI agents with built-in SLA monitoring? → AfrexAI handles deployment, monitoring, and maintenance for $1,500/mo → Book a call: https://calendly.com/cbeckford-afrexai/30min → Learn more: https://afrexai-cto.github.io/aaas/landing.html