Run the War Room, don't just read about it
The reps startups actually run: drop into a live production incident and drive it end to end — triage, RCA, safe fix, prevention and post-mortem — graded like a staff SRE.
How a real incident is actually run
Not a quiz about incidents — the full loop a senior engineer drives when production is on fire.
Triage & impact
Assess severity and blast radius, decide mitigate-vs-investigate, and page the right people — under a ticking clock.
Investigate
Pull logs, metrics, recent deploys, traces and config from an AI incident channel that reveals only what you ask for — it never hands you the answer.
Ship a safe fix
Choose and justify the fix that stops the bleeding without making it worse — rollback, flag, hotfix or scale — and defend the trade-off.
Architect prevention
Design the durable change so this class of failure can't recur: the missing guardrail, test, alert or circuit breaker.
Blameless post-mortem
Write the timeline, contributing factors and action items the way a healthy engineering org actually does.
As close to the real thing as it gets
Investigation you have to drive, grading that holds a real bar, and incidents drawn from history.
An AI incident channel
A realistic on-call channel role-plays the incident, revealing evidence only when you ask the right question — so you practise investigation, not guessing.
Graded like a staff SRE
Scored across triage, investigation method, correct root cause, safe fix, durable prevention and communication — with specific feedback.
Original + real incidents
Drills span classic failure modes and real documented outages (GitLab, Knight Capital, Cloudflare, S3) with variations to reason through.
Learn from 35 real, cited disasters
Every drill is paired with a browsable library of real documented failures — Knight Capital's $440M/45min, the AWS S3 typo, Cloudflare's regex CPU meltdown, Equifax — each mapped to the principle it teaches and the cost of ignoring it. Feeds into results that raise your verified competencies and credential.
Feeds your credential
Incident scores raise verified competencies and count toward mastery-gated advancement.
Principle-first
Each failure teaches a concrete concept — idempotency, circuit breakers, rate limits, rollback.