Incident Commander

Incident Commander Guide

Lead incident response with structured communication, clear ownership, and systematic resolution.

Incident Commander Role

The IC (Incident Commander) is responsible for:

• Coordination: Ensuring right people are engaged
• Communication: Keeping stakeholders informed
• Decision-making: Making calls when consensus isn't possible
• Documentation: Ensuring timeline is captured

The IC does NOT need to be the person fixing the problem.

Incident Severity Levels

| Level | Description | Response Time | Examples | |-------|-------------|---------------|----------| | SEV1 | Critical - Complete outage | Immediate | Total service down, data loss, security breach | | SEV2 | Major - Significant impact | 15 min | Core feature broken, major degradation | | SEV3 | Minor - Limited impact | 1 hour | Non-critical feature down, workaround exists | | SEV4 | Low - Minimal impact | Best effort | Cosmetic, single user affected |

Incident Response Workflow

Phase 1: Detection & Triage (0-5 minutes)

1. Acknowledge the alert
2. Quick assessment:
   - What's broken?
   - Who's affected?
   - What's the blast radius?
3. Assign severity level
4. Declare incident if SEV1/SEV2

Phase 2: Mobilization (5-15 minutes)

1. Create incident channel: #inc-YYYYMMDD-[brief-description]
2. Post initial summary (template below)
3. Page relevant teams if needed
4. Assign roles:
   - IC (Incident Commander) - you or delegate
   - Tech Lead - driving investigation
   - Comms Lead - external communication

Initial Incident Post Template:

🚨 INCIDENT DECLARED

**Severity:** SEV-[X]
**Status:** Investigating
**Impact:** [Who/what is affected]
**Started:** [Time] UTC

**Current Understanding:**
[Brief description of symptoms]

**Roles:**
- IC: @[name]
- Tech Lead: @[name]
- Comms: @[name]

**Next Update:** [Time] (every 15-30 min for SEV1/2)

Phase 3: Investigation (Ongoing)

1. Gather data:
   - Recent deployments?
   - Configuration changes?
   - External dependency issues?
   - Error patterns in logs?
   - Metrics anomalies?

2. Form hypothesis and test

3. Identify mitigation options:
   - Can we rollback?
   - Can we scale?
   - Can we failover?
   - Do we need a hotfix?

Phase 4: Mitigation

1. Choose mitigation approach
2. Communicate plan before executing
3. Execute with verification at each step
4. Monitor for improvement
5. Confirm resolution

Phase 5: Resolution

1. Verify service is healthy
2. Update status page
3. Send resolution communication
4. Create postmortem ticket
5. Schedule postmortem meeting (within 48h for SEV1/2)

Communication Templates

Status Update (Every 15-30 min)

📊 INCIDENT UPDATE - [Time] UTC

**Status:** [Investigating/Identified/Mitigating/Resolved]
**Impact:** [Current impact]

**Update:**
[What we've learned, what we're doing]

**Next Steps:**
[What's happening next]

**Next Update:** [Time] UTC

Resolution Communication

✅ INCIDENT RESOLVED - [Time] UTC

**Duration:** [X hours Y minutes]
**Root Cause:** [Brief description]
**Resolution:** [What fixed it]

**Impact Summary:**
- Users affected: [number]
- Duration: [time]
- SLA impact: [yes/no]

**Next Steps:**
- Postmortem scheduled: [date/time]
- Postmortem doc: [link]

Thank you to everyone who helped respond.

Best Practices

DO

• Keep incident channel focused on resolution
• Use threads for side discussions
• Update status page early and often
• Make decisions when stuck (bias for action)
• Rotate IC if incident is long (>4 hours)
• Take breaks - fatigue causes mistakes

DON'T

• Blame individuals
• Make changes without communicating
• Forget to update stakeholders
• Skip the postmortem
• Let scope creep into fixing unrelated issues

Role Cheat Sheet

| Role | Responsibility | Who | |------|----------------|-----| | IC | Coordination, decisions, communication | Declared or on-call | | Tech Lead | Investigation, fix implementation | SME for affected service | | Comms Lead | Status page, customer comms | Support/Comms team | | Scribe | Document timeline | Anyone available | | Subject Matter Experts | Deep knowledge | Paged as needed |

Escalation Triggers

Escalate to leadership when:

• SEV1 lasting >30 minutes
• Data breach suspected
• Customer SLA breach imminent
• Media attention expected
• Regulatory notification required

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: latestaiagents
• Source: latestaiagents/agent-skills
• License: MIT
• Homepage: https://latestaiagents.com

Install and usage instructions live in the source repository linked above.