| name | incident-management |
| description | Cross-platform incident management router. Detects the alerting/on-call platform in use (Grafana IRM, incident.io, Better Stack, Datadog, PagerDuty) and routes to the appropriate specialist skill. Handles platform-agnostic tasks — alert design, on-call setup, runbook authoring, and blameless postmortems — without requiring platform knowledge. |
| context | fork |
| allowed-tools | Bash(grep:*), Bash(cat:*), Bash(ls:*), Bash(git diff:*), Bash(git log:*), Bash(git status), Glob, Grep, Read, WebFetch |
| tags | ["incident-response"] |
Incident Management
Cross-platform incident management entry point. Detect the project's platform, then route to the appropriate specialist or execute a platform-agnostic task.
When to Use
- Setting up alerting or on-call for a project without knowing which platform to use
- Designing alert rules, severity levels, or escalation policies
- Creating or improving runbooks
- Running a blameless postmortem
- Migrating from OpsGenie (shutting down April 2027) to a modern platform
- Getting guidance when the team is evaluating incident management platforms
Platform Detection
Scan the repository to identify the active platform before routing:
!`grep -r "grafana_oncall\|grafana_irm\|grafana-irm" . --include="*.tf" --include="*.hcl" -l 2>/dev/null | head -5`
!`grep -r "incident\.io\|incidentio\|provider.*incident" . --include="*.tf" --include="*.hcl" -l 2>/dev/null | head -5`
!`grep -r "betterstack\|better_uptime\|better-stack" . --include="*.tf" --include="*.hcl" -l 2>/dev/null | head -5`
!`grep -r "pagerduty\|pagerduty_" . --include="*.tf" --include="*.hcl" -l 2>/dev/null | head -5`
!`grep -r "datadog_monitor\|DD_API_KEY\|datadoghq" . --include="*.tf" --include="*.hcl" --include="*.env*" -l 2>/dev/null | head -5`
Routing Logic
| Platform Detected | Route To |
|---|
grafana_oncall_* or grafana_irm_* in Terraform | /grafana-irm specialist |
provider "incident" or incidentio_* in Terraform | /incident-io specialist |
betterstack_* or better_uptime_* in Terraform | /better-stack specialist |
pagerduty_* in Terraform | /pagerduty specialist (if available) or generic guidance |
datadog_monitor or DD_API_KEY | /observability skill with incident mode |
| No platform detected | Present evaluation matrix below, ask user to confirm |
Platform-Agnostic Modes
When invoked without a detected platform, or explicitly for cross-cutting concerns:
design-alerts
Design symptom-based alert rules from scratch. Use references/alerting-patterns.md and references/severity-taxonomy.md.
Gather:
- Service name and language/framework
- Key user-visible operations (checkout, login, search, etc.)
- Current p50/p75/p99 latency baselines (if known)
- Existing monitoring stack (Prometheus, Datadog, Grafana, etc.)
Output: severity-classified alert definitions with thresholds, burn-rate windows, and runbook URLs.
on-call-setup
Design an on-call rotation structure. Use references/on-call-rotation.md.
Gather:
- Number of engineers available for rotation
- Team timezone distribution
- Current shift length preference
- Whether a shadow/mentorship rotation is needed
Output: rotation schedule template, escalation policy tiers, handoff checklist.
runbook
Create or improve an alert runbook. Use references/runbook-template.md.
Gather:
- Alert name and firing condition
- Service affected and owner team
- Known common causes
- Existing remediation steps (if any)
Output: complete runbook following the standard six-section structure.
postmortem
Facilitate a blameless postmortem. Use references/postmortem-template.md.
Gather:
- Incident timeline (start, detection, triage, resolution)
- SEV level and user impact
- Immediate contributing factors
- Systems and people involved
Output: completed postmortem document with action items.
evaluate
Platform evaluation for teams choosing between Grafana IRM, incident.io, Better Stack, and PagerDuty.
Present the comparison matrix from references/alerting-patterns.md and guide through:
- Existing stack compatibility (Grafana, OTel, Slack, Datadog)
- MCP integration requirements
- Terraform workflow maturity
- Budget and seat count constraints
Platform Status Warnings
Surface these when relevant:
- OpsGenie: New sales ended June 2025; full shutdown April 5, 2027. Recommend migration to Grafana IRM, incident.io, or Better Stack.
- Grafana OnCall OSS: Maintenance mode March 2025; archived March 2026. Route all OSS OnCall users to Grafana Cloud IRM.
- FireHydrant: Acquired by Freshworks 2025. Avoid new skill investment.
- Squadcast: Acquired by SolarWinds 2025. Avoid new skill investment.
References
references/alerting-patterns.md — Platform-agnostic alert design, SLO burn rate, fatigue mitigation
references/severity-taxonomy.md — SEV0–SEV3 definitions, escalation policy tiers
references/on-call-rotation.md — Rotation patterns, handoff templates, follow-the-sun
references/runbook-template.md — Standard runbook structure
references/postmortem-template.md — Blameless postmortem structure and facilitation guide