Nightly Saturation Sweep with Approval-Gated Scaling

Proactively heads off capacity incidents instead of waiting to be paged. On a schedule, the agent reads Datadog trends for disk, memory, connection pools, and queue depth, projects which resources will hit their limits soon, and drafts preemptive remediation. The operator approves before any scaling or cleanup runs.

Use it when slow-burn saturation issues (disks filling, pools exhausting) repeatedly turn into 3 a.m. pages. Running it nightly converts those into a calm, batched morning decision with a clear proposed action.

1. 1A nightly schedule triggers the flow.
2. 2The agent queries Datadog for capacity and utilization trends across watched resources.
3. 3A logic step flags resources whose projected trajectory crosses a saturation threshold.
4. 4For each flagged resource the agent drafts a specific scaling or cleanup action with the command and expected headroom gained.
5. 5The proposals are posted to Slack as a digest with per-item Approve controls.
6. 6On approval, the shell action executes the chosen items and reports the new utilization.