Scheduling & Triggers
Stop waiting for alerts. Hyground executes operational workflows on a schedule: daily health checks at 7:55am, nightly security scans, weekly capacity reviews every Monday. It collects evidence and takes action before issues escalate into incidents.
Every day at 7:55am
Before anyone opens their laptop, Hyground has already swept your systems for overnight anomalies, error rate spikes, disk pressure, failed cron jobs, unexpected pod restarts. Evidence is collected, findings are structured, and the report lands in Slack before standup.
Every night at 2am
Every night, deeply analyse the trivy findings from the CRDs with your production code. What is a critical issue for your application? What needs immediate fixing? Triage and consolidate findings.
Every Monday morning
A weekly operations digest lands in the engineering manager's inbox. Services approaching saturation, storage growth projections, cost anomalies, and recommended actions, with supporting evidence from the past seven days.
Schedules wrap Skills
A Skill is a reusable procedure your team writes once: an investigation prompt plus the tools it can use. A schedule wraps that Skill with a cadence and the prompt that invokes it. Replace "configure a cron job in YAML" with "pick a Skill, set a cadence, save."
Define reusable operational procedures once and invoke them from a schedule, an event, or an ad-hoc prompt.
Choose a built-in skill or write a custom one. Specify what to check, what thresholds matter, what actions to take, and what evidence to include in findings.
Configure the schedule in the web UI, or ask the agent to create it for you in chat.
Results are always stored in Hyground for review. Findings are structured, evidence-backed, and include any actions taken or recommended. Export to Slack, Microsoft Teams, or email is optional.
A schedule is a small, explicit object. No buried config, no surprise auth context.
A human-readable identifier shown in the dispatch log and the agent's chat replies.
The exact request sent to the agent each run. Same shape as an ad-hoc question.
A cron expression or a one-off datetime. Standard 5-field syntax, supports timezone.
Toggle a schedule off without deleting it. Disabled schedules stay in your library and can be re-enabled later.
From chat
When a request sounds recurring ("send me this every Monday"), the agent offers to schedule it. It confirms cadence and the first-run time in your timezone before anything is saved.
Reliable
Multiple replicas claim due schedules with row-level locking, so no run fires twice. Runs missed during a crash recover automatically when the dispatcher restarts.
Auditable
Each scheduled run opens a DispatchTask with a clickable URL. The session holds the transcript, tool calls, files retrieved, and final reports. That is the evidence.
Findings always land in Hyground. Optional exports push them to the channel your team already uses.
Post findings to a channel, DM the on-call, or create a thread per workflow execution.
Send findings as cards to a Teams channel. Works with Teams notifications and workflows.
Deliver a formatted operations summary to individuals or distribution lists on a schedule.
See scheduled operations running against your own infrastructure. We will set up a morning health check and show you what evidence-backed findings look like in Slack.
Check out our sandbox or schedule a demo with our team and experience sovereign AI for DevOps firsthand.