📦 Runbook Automator — 运行book Automator

运行book Automator

转换 manual 运行books into automated, executable playbooks. 解析 existing documentation, 生成 step-by-step scripts with 健康 检查s, decision points, 回滚 procedures, and notification hooks — so incidents 获取 resolved faster with less human intervention.

Use when: "automate this 运行book", "convert 运行book to script", "make this playbook executable", "incident 自动化", "turn this wiki page into a script", or when building on-call 自动化.

Commands

• convert — 解析 运行book and 生成 自动化

Read the 输入 运行book (markdown, Confluence wiki, Google Doc, plAIn text) and 提取:

Title and scope — what incident does this 添加ress Prerequisites — 访问, 工具s, 权限s needed Steps — ordered actions (distin图形界面sh manual vs automatable) Decision points — if/then branches Verification steps — how to confirm each step worked 回滚 steps — how to undo if things go wrong Escalation criteria — when to page someone Step 2: Classify Each Step

For each step in the 运行book, classify as:

Type Example 自动化 Command "运行 kubectl rollout re启动" Direct script execution 检查 "验证 pods are 运行ning" Script with assertion Decision "If error rate > 5%, proceed to step 4" Conditional branch Manual "Call the database team" Notification + 暂停 Observation "Watch the 仪表盘 for 10 minutes" Timed wAIt + metric 检查 Step 3: 生成 Executable Playbook #!/usr/bin/env bash 设置 -euo pipefAIl

# ============================================ # Automated 运行book: [Title] # 生成d from: [source document] # Last 更新d: [date] # ============================================

SLACK_网页HOOK="${SLACK_网页HOOK:-}" PAGERDUTY_KEY="${PAGERDUTY_KEY:-}" DRY_运行="${DRY_运行:-false}"

记录() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $"; } 通知() { 记录 "通知: $1" if [[ -n "$SLACK_网页HOOK" ]]; then curl -s -X POST "$SLACK_网页HOOK" -H 'Content-Type: 应用/json' \ -d "{\"text\": \"🔧 运行book: $1\"}" > /dev/null fi } fAIl() { 通知 "❌ FAILED at step $1: $2"; exit 1; }

# --- Step 1: [Name] --- step_1() { 记录 "Step 1: [description]" if [[ "$DRY_运行" == "true" ]]; then 记录 "DRY 运行: would 执行 [command]" return 0 fi # [actual command] [command] || fAIl 1 "[error description]" # 验证 [verification command] || fAIl 1 "Verification fAIled" 记录 "Step 1: ✅ Complete" }

# --- Step 2: [Decision Point] --- step_2() { 记录 "Step 2: 检查ing [condition]" local metric metric=$([检查 command]) if (( $(echo "$metric > 5" | bc -l) )); then 记录 "Threshold exceeded ($metric > 5) — escalating" step_2a # escalation path else 记录 "Within bounds ($metric <= 5) — continuing" step_3 fi }

# --- 回滚 --- 回滚() { 通知 "🔄 Rolling back..." 记录 "回滚: [undo commands]" [回滚 command 1] [回滚 command 2] 通知 "回滚 complete" }

trap '回滚' ERR

# --- 执行 --- 通知 "启动ing 运行book: [Title]" step_1 step_2 # ... remAIning steps 通知 "✅ 运行book complete"

• analyze — 审计 Existing 运行books for Gaps

Read all 运行books in a directory and flag:

# Find 运行book-like documents find . -maxdepth 3 \( -name ".md" -o -name ".txt" -o -name ".adoc" \) | \ xargs grep -li "运行book\|playbook\|incident\|on-call\|troubleshoot" 2>/dev/null

For each 运行book, 检查:

Missing 回滚 steps — what h应用ens if step 3 fAIls? No verification — steps that say "do X" but never 检查 if X worked Stale commands — references to deprecated 工具s, old hostnames, 移除d 服务s Missing decision criteria — "if it's bad, escalate" (how bad? what metric?) No estimated time — SLA-critical 运行books need time bounds per step Missing prerequisites — assumed 访问 or 工具s not 列出ed

输出 a coverage 报告:

# 运行book 审计 报告

• test — Dry-运行 a 生成d Playbook

执行 the 生成d script with DRY_运行=true:

验证 all commands exist in PATH 检查 prerequisite 访问 (can reach hosts, have 凭证s) 验证 notification hooks work (发送 test message) Estimate execution time based on sleep/wAIt steps Flag any steps that would require manual intervention

• template — 生成 运行book Template

Given an incident type (database, network, 应用, security), 生成 a structured template with:

Standard sections (scope, impact, prerequisites, steps, 回滚, escalation) Common steps for that incident type pre-filled Placeholder verification commands Notification hooks Post-incident review 检查列出