Sre foundations
Applies Google SRE principles to production systems — Production Readiness Reviews, TOIL reduction, incident lifecycle, and blameless postmortems. Use when preparing a service for production, evaluating operational maturity, responding to incidents, or writing postmortems. Also use when someone asks about on-call, error budgets, SRE engagement, or reducing manual operational work.From its SKILL.md
npx -y skills add valdomirosouza/agent-skills --skill sre-foundationsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
2.7 KB, 601 tokens by cl100k_base, as published. Nobody here has run it
SRE Foundations
Core principle
SRE treats operations as a software problem. Reliability is a feature. Every manual, repetitive task is a bug to eliminate.
Contents
- PRR (Production Readiness Review): prr-checklist.md
- TOIL identification and elimination: See below
- Incident lifecycle and severities: incident-response.md
- Blameless postmortem template: incident-response.md
TOIL — Identification and Policy
Work is TOIL if it is: manual, repetitive, automatable, reactive, and has no enduring value.
| Metric | Target |
|---|---|
| Max TOIL as % of eng time | ≤ 50% |
| Sustainable target | ≤ 25% |
| Any task taking > Xh/week | Must have automation backlog item |
Automation categories:
| Category | Tools |
|---|---|
| Deployment | ArgoCD, Spinnaker, Harness CD |
| Scaling | HPA, VPA, KEDA |
| Remediation | Circuit breaker, auto-restart, auto-healing |
| Alerting | AlertManager, PagerDuty, OpsGenie |
| Incident | FireHydrant, Incident.io, PagerDuty Rundeck |
Error Budget Policy
Error Budget = 1 - SLO target
Example: SLO 99.9% → Error Budget = 0.1% of time in the period
Budget < 50% → Feature freeze; focus on reliability
Budget < 10% → Deploy freeze (except hotfixes); incident review mandatory
Budget = 0 → Full deploy freeze; executive escalation; SLO review
PRR Gate
Every service must pass PRR before production. See prr-checklist.md for the full checklist.
PRR prerequisites (must complete first):
- NALSD design validated (section 4.0 of main spec)
- SLI/SLO defined and documented
- Runbook written and reviewed by someone external to the team
- Chaos experiments planned
Incident Severity Matrix
| SEV | Criterion | Response SLA | Lead |
|---|---|---|---|
| SEV-1 | Production down, massive user impact | < 5 min | Senior on-call + CTO |
| SEV-2 | Severe degradation, critical feature affected | < 15 min | On-call |
| SEV-3 | Partial degradation, workaround available | < 1h | Responsible team |
| SEV-4 | Minimal impact, no user effect | < 1 business day | Backlog |
Full incident lifecycle and postmortem template → incident-response.md
What ships with it: 2 files
5.2 KB alongside SKILL.md
- incident-response.md2.6 KB
- prr-checklist.md2.6 KB