DNS
TXT record
Add a one-time token to your domain's DNS. We check it, then keep checking.
⚠ How it works
Deploy a real GameDay in four steps
Verify ownership, pick a scenario, deploy one agent, break things on purpose then read the report. No platform to build first.
Step 1 Prove it's yours
Nothing runs until you've shown us you own the target. This is the gate, not a checkbox.
Cloud
Scoped IAM role
Grant a narrow, read-and-inject role in your AWS, Azure or GCP account. Revoke it any time.
Agent
Agent token
The agent runs inside your network and registers with a signed token tied to your environment.
Always
Re-verified
Verification expires. If it lapses, scheduled runs pause automatically.
Step 2 Pick your disaster
Choose from the scenario library region outage, fibre cut, third-party API meltdown or compose your own.
Blast radius
Limit a run to one service, one zone, or a percentage of instances.
Duration
Every run has a hard time limit. When it's up, faults roll back.
Hypothesis
Write down what you expect to happen. The report grades you against it.
Abort conditions
Tie runs to your SLOs. Error rate crosses the line? We stop, instantly.
Step 3 Deploy with one command
The agent is outbound-only. No inbound ports, no VPN, no firewall tickets.
docker run -d --name omgdaas-agent \ -e OMG_TOKEN=$YOUR_ENV_TOKEN \ omgdaas/agent:latest # or Kubernetes helm install omgdaas omgdaas/agent --set token=$YOUR_ENV_TOKEN
Step 4 Run, watch, abort
Live timeline
Every fault injected, every alert fired, every human reaction on one screen.
The big red button
Anyone on the run can abort. Faults roll back in seconds.
The report
Hypothesis vs reality, findings, action items you can push to Jira, and a PDF for your auditor.
Do it again
Schedule it monthly. Resilience is a habit, not a project.
Plugs into what you already use
Slack and PagerDuty for alerts and run updates. Datadog and Grafana for steady-state checks and abort signals. Jira for follow-ups. Webhooks and a REST API for everything else.