Reliability
Overview
Three leaves operate a service in production: instrument it so questions have answers, set the reliability target and alerting on it, and run the outage when it comes. Instrument, then target, then respond.
When to use
- Adding structured logs, metrics or traces to a service; choosing what to instrument.
- Writing an SLI/SLO, an error-budget policy, burn-rate alert rules, RTO/RPO per tier.
- Declaring and running an outage; status communications; the postmortem.
When not to use
- Canary ramps, rollback mechanics or release staging — outside this skill.
- Latency budgets, profiling, load tests — outside this skill.
- Browser-side Core Web Vitals — outside this skill.
Precedence
The deepest loaded instruction governs its uniquely-claimed scope; every
scope:string is claimed by exactly one file in this tree; no leaf may widen a prohibition declared by an ancestor — leaves narrow, never override upward.
Subskill index
| Leaf | Task | Trigger |
|---|---|---|
subskills/observability.md |
Logs, metrics, traces | "add tracing" |
subskills/slo-reliability.md |
SLOs, error budgets, alerts, DR | "set an SLO" |
subskills/incident-response.md |
Run an outage, postmortem | "prod is down" |
See also
subskills/observability.md— see the system.subskills/slo-reliability.md— set the bar.subskills/incident-response.md— hold the bar when it breaks.