vercel-incident-runbook

name	vercel-incident-runbook
description	Vercel incident response procedures with triage, instant rollback, and postmortem. Use when responding to Vercel-related outages, investigating production errors, or running post-incident reviews for deployment failures. Trigger with phrases like "vercel incident", "vercel outage", "vercel down", "vercel on-call", "vercel emergency", "vercel broken".
allowed-tools	Read, Grep, Bash(vercel:*), Bash(curl:*)
version	1.18.0
license	MIT
author	Jeremy Longshore <[e-mail]>
tags	["saas","vercel","incident-response","runbook"]
compatibility	Designed for Claude Code, also compatible with Codex and OpenClaw

Vercel Incident Runbook

Overview

Step-by-step incident response for Vercel deployment failures, function errors, and platform outages. Covers rapid triage, instant rollback, communication templates, and postmortem procedures.

Prerequisites

• Access to Vercel dashboard and CLI
• Access to Vercel status page (vercel-status.com)
• Communication channels (Slack, PagerDuty) configured
• Log drain or runtime log access

Instructions

Step 1: Rapid Triage (First 5 Minutes)

# 1. Check if it's a Vercel platform issue
curl -s "https://www.vercel-status.com/api/v2/summary.json" \
  | jq '.status.description, [.components[] | select(.status != "operational") | {name, status}]'

# 2. Check current production deployment status
vercel ls --prod
vercel inspect $(vercel ls --prod --json | jq -r '.[0].url')

# 3. Check recent deployments — did a deploy just happen?
curl -s -H "Authorization: Bearer $VERCEL_TOKEN" \
  "https://api.vercel.com/v6/deployments?target=production&limit=5&projectId=prj_xxx" \
  | jq '.deployments[] | {uid, state, createdAt: (.createdAt/1000 | todate), url}'

# 4. Check function logs for errors
vercel logs $(vercel ls --prod --json | jq -r '.[0].url') --level=error --limit=20

Step 2: Decision Tree

Is vercel-status.com showing an incident?
├── YES → Vercel platform issue
│   ├── Subscribe to updates on status page
│   ├── Post internal status: "Vercel platform incident — monitoring"
│   └── No action needed from us — wait for Vercel resolution
│
└── NO → Issue is in our deployment
    ├── Did a deployment happen in the last 30 minutes?
    │   ├── YES → Likely deployment regression
    │   │   └── ROLLBACK immediately (Step 3)
    │   └── NO → Application-level issue
    │       ├── Check function logs for new errors
    │       ├── Check external dependency status (DB, APIs)
    │       └── Investigate and hotfix (Step 4)
    │
    └── Is the issue region-specific?
        ├── YES → Check function regions, possible edge issue
        └── NO → Global issue, check code and env vars

Step 3: Instant Rollback (< 30 Seconds)