Operational tasks
Catch degradation before it pages anyone.
Proactive health checks, deploy watches and readiness reviews, running against a live map of your production
Workflows
Automate any recurring job on a schedule or a trigger.
Run your workflows on a cadence, or fire agents on a deploy, an alert, or an investigation closing
Your workflows and skills
Customise agents for your operational work.
Health checks, deploy watches and cost reviews, written in plain language and started from a skill in the catalog
Priority
Start the day knowing where you need to focus.
Always-on workflows investigate first, so you open the day on verified findings and what could not be checked
The warning
The warning arrives before your alerting does.
What is degrading, the deploy that correlates, what sits downstream, and what could not be checked
We jumped between 5 dashboards to see what broke. Now on-call sees the cause and the impact, and they fix it.
Frequently asked questions.
What can I leave running?
Eight predefined skills: health check, post-deploy, pre-merge, alert triage, readiness review, and the rest. Or a skill you write, such as an hourly SLO check on customer-facing services in the prod-us-east-1 environment.
How does it know what to watch?
It reads the map: what is running, who owns it, what just shipped, what is missing. Same record as investigations.
How do I start one?
Alert triage is already on. Add a predefined skill, or write your own. The trigger is on the card. You can also run one from chat.
What triggers are supported?
A schedule, a GitHub event, a Slack alert, or an investigation closing.
Does it block a deploy?
No. Pre-merge ranks the change against the map. A person still ships.
Does it take the shift?
No. A person still decides.
What if it is wrong?
Disable it. Every run is recorded, including runs that found nothing.
Can a run open an investigation?
Yes. Open the investigation from the run. You can also keep asking in chat.