Respond to incidents
without the scramble.
Go from triggered to resolved on one screen. Triage with AI, escalate, and pull in whoever you need.
Know what you’re dealing with.
Spike titles it, suggests severity and priority, and shows how it behaved last time.
Incident title
Checkout CPU pinned at 98% on prod-web-3
Written by AIDetected from payload
Detected from payload
Added from your escalation policy
A reminder pings before the timeout — so nothing sits acknowledged and forgotten.
By a person
Someone marks it resolved.
Auto-resolved
A recovery payload closes it — Grafana back to OK.
By a timer
Auto-resolves after a set window: 2h, 3d, your call.
AI severity & priority
Spike reads the alert and recommends both, with a one-line why.
Root-cause hints
Ask over MCP and Spike points at likely causes from your incident history.
It’s happened before
Past resolution notes, suppression, and repeat counts resurface on the incident.
Put an agent on every incident.
Spike speaks MCP. Connect your AI client or your own agents to triage and act on incidents.
On-call CopilotMCP
Your agent
Reading INC-2481 and deciding what to do…
spike.get_incident("INC-2481")
SEV1 · checkout-apispike.set_priority("P1")
Updatedspike.acknowledge()
Maya pagedspike.escalate("secondary")
NotifiedRuns on every incident.triggered
Ask from Slack
@Spike answers about any incident, right in the channel.
Chat over MCP
Connect Claude, OpenAI, or Perplexity and ask about incidents in plain English.
API & webhooks
Drive the same triage and actions straight from your own code.
Every response action, one incident.
Escalate, run a script, open a war room, or sync a ticket. All without leaving Spike.
RUN A SCRIPT
POST A STATUSPAGE UPDATE
Elevated API error rates
Resolved- ResolvedAug 27, 2026 · 14:20
Error rates are back to normal and the fix has held for 30 minutes. We’re marking this resolved.
- MonitoringAug 27, 2026 · 13:45
A fix has been deployed and error rates are falling. We’re watching to confirm full recovery.
- IdentifiedAug 27, 2026 · 13:12
The issue is a saturated database connection pool. We’re rolling out increased capacity now.
- InvestigatingAug 27, 2026 · 12:58
We’re seeing elevated error rates on the API and are investigating the cause.
Escalate anywhere
One tap moves the incident to the next on-call, over Slack, Teams, phone, and SMS.
Run a script
Trigger a GitHub workflow or any saved webhook, straight from the incident.
Spin up a war room
Pull responders into a call and Spike tells everyone where to go.
Post a statuspage update
Tell customers on your public page, or clear it with stakeholders internally.
Sync a ticket
Push to Linear, Jira, or ClickUp so the owning team picks it up.
Loop in support
Bring Zendesk, Freshdesk, or SupportPal in when customers are hit.
A room to work the incident.
Spike posts the incident to Slack and can open a channel to work it, right from the thread.
Spike created this private channel for INC-2481
SpikeAPP20:05
Checkout CPU pinned at 98% on prod-web-3
Incident INC-2481 · checkout-api
Close the loop, keep the learning.
Write a resolution note and it sticks to the incident, so the next person knows how it was fixed.
Maya Khan
@Jordan is this the connection-pool bug from March again?
Jordan Reid
Yep. Bumped pgbouncer to 40 and restarted web-3.
Raised pgbouncer pool 20 → 40. Root cause: N+1 query in /cart. Watch pool saturation on deploys.
Resurfaces automatically the next time this fires.
Summary
Checkout latency spiked when prod-web-3 held 98% CPU for ten minutes. Paged at 20:05, acknowledged in 1m 12s, resolved in 8m.
Timeline
- 20:05 — Triggered, SEV1
- 20:06 — Maya acknowledged
- 20:11 — Deploy freeze applied
- 20:13 — Resolved
Impact
Elevated checkout p99 for ~8 minutes. No orders lost.
Root cause
An N+1 query in /cart under load pinned the pgbouncer pool.
Action items
- Cap the pool and add a query budget
- Alert on p99 before saturation
Stop firefighting.
Start responding.
Give your on-call one place to triage, act, and close every incident.