| id | triage-cost-spike |
| name | Triage Cloud Cost Spike |
| description | Identify the primary drivers of a sudden cloud cost increase and implement safe mitigations. |
| tags | ["cost","finops","triage","oncall"] |
| maturity | draft |
| inputs | {"cloud":"string?","account_or_project":"string?","time_window":"string?"} |
| outputs | ["top_cost_drivers","suspected_root_causes","mitigation_options"] |
| tools_allowed | ["billing.read","metrics.query"] |
| safety | {"default_mode":"read_only","forbidden":["billing.write"],"requires_confirmation_for":["billing.write"]} |
When to use
Use when daily cost increases significantly versus baseline.
Preconditions
- You can access billing/cost breakdown and recent change/deploy history.
Procedure
- Confirm the spike magnitude and when it started.
- Break down by service, region, and top resources/tags.
- Identify the change driver: deploy, scaling, traffic, misconfiguration, abuse.
- Validate whether spend is expected (planned test) or unexpected.
- Choose the safest mitigation: pause non-critical workloads, cap scaling, fix loops.
Decision points
- Compute spike: check autoscaling, instance sizes, and runaway jobs.
- Network spike: check egress, NAT, CDN settings, cross-region traffic.
- Storage spike: check log retention, snapshots, object lifecycle policies.
Verification
- Cost trend returns toward baseline within the next billing window.
- The change driver is addressed (scaling capped, job stopped, retention fixed).
Rollback / undo
- Revert cost mitigation that impacts critical reliability.
- Restore paused workloads in a controlled manner.
Escalation
- FinOps/cloud owners for billing anomalies or reserved commitment adjustments.
- Security team if spike may be caused by abuse/compromise.
Examples
Start with a service-by-service cost breakdown, then drill into the top 1-3 line items.