SLO and Alert Review
Review existing SLOs and burn-rate alerts: fix PromQL math, noisy or silent pages, low-traffic traps, and targets that miss reality.
Before & after
Same prompt, same agent — the only difference is the skill.
Defining SLOs is important for reliability. You'll want to think about what uptime or latency targets make sense for your service. Typically services aim for 99.9% or 99.95% availability. You could use error budgets to decide when to focus on reliability versus shipping features. It's also good to define alerts so you know when you're at risk of missing your targets. What level of reliability does your service need? Do you have historical performance data I could use to set realistic targets?
SLO FRAMEWORK: API Infrastructure SERVICE: Authentication API (auth.example.com) OBJECTIVE 1: Availability Target: 99.95% uptime (21.6 minutes downtime/month) Error budget: 3.15 hours/month OBJECTIVE 2: Latency (p99) Target: <200ms for token generation Error budget: 3 hours/month over p99 SLO OBJECTIVE 3: Error Rate Target: <0.1% 5xx errors Error budget: 72 errors/day on 1M daily requests BURN RATE ALERTS: • Fast burn (6h window): If consuming >50% budget in 6 hours → alert ops team • Slow burn (30d window): If consuming >100% budget → freeze feature launches • Critical: If availability drops below 99% in any day → incident response ERROR BUDGET ALLOCATION (3.15 hrs/month): • Deployments: 1.5 hrs (1 deploy/week, accept brief latency spike) • Database maintenance: 1 hr (monthly patches) • Reserved: 0.65 hrs (unexpected incidents) ACTION GATES: ✓ 75% consumed this month → no risky deployments in last week ✓ 100% consumed → rollback last 2 changes, postmortem required METRICS (tracked in Prometheus/Datadog): • Availability: (success requests / total requests) × 100 • Latency p99: histogram quantile(0.99, response_time_seconds) • Error rate: rate of 5xx errors per 1-minute window
About this skill
Teams with SLOs still get paged for nothing and miss real outages, usually because of a few precise errors in queries and alert rules. This skill reviews what you run today and returns corrected rules.
What it does
- Checks query math: ratio of sums, no averaged percentiles, consistent filters, bucket boundaries at the latency threshold
- Verifies multiwindow multi-burn-rate thresholds with the arithmetic shown
- Finds silent failure modes: no-traffic outages, measurement points that cannot see errors, mismatched windows
- Handles low-traffic services and demotes cause-based pages
- Compares targets with achieved history and missed incidents
What you get
Ranked findings with evidence, corrected rules ready to paste, recommended target changes with reasons, and an alert-noise summary.
Best for
SRE and platform teams on Prometheus, Datadog, Grafana, CloudWatch, or SLO tools whose on-call is noisy or missed an incident.
SKILL.md preview
---
name: slo-alert-review
description: Review and tune SLOs, SLI queries, error budget policies, and burn-rate alert rules that already exist - Prometheus/PromQL, Datadog, Grafana, CloudWatch, or Nobl9/Sloth/OpenSLO configs - finding math errors, blind spots, noisy or silent alerts, and targets that no longer match reality, then return corrected rules and thresholds. Use when the user says "review our SLO alerts", "we get paged too much", "our SLO never fires", "check this PromQL burn rate rule", "are our SLO targets right?", "tune our error budget alerts", "we missed an outage and the SLO stayed green", or pastes alert rules or an SLO config.
---
# SLO and Alert Review
… (sign up to view the full skill)More development skills
View all Development skills →Chaos Experiment and Game Day Planner
Plans chaos experiments and game days: hypothesis, blast radius, abort conditions, tool manifests, rollback, and post-run review.
Claude Code Cost and Latency Routing
Configure Claude Code Router with model tiers, explicit fallbacks, compatibility checks, and a cost-aware validation plan.
CI/CD Pipeline Builder with Gated Deploys
Generates CI/CD workflows from your repo's real stack: cached fast-fail checks, pinned actions, and staging-to-prod deploys with gates.
Terraform Modules and Safe Deployments
Get reusable Terraform modules, safe state migrations, explicit provider wiring, and CI gates tied to reviewed plans.