reviewing-release-runbook

Reviews release runbooks against an 11-condition executability and safety bar, emitting an approve or revise verdict.

1|1|Updated May 24, 2026
One-click install
npx skills add https://github.com/bm629/agent-skills --skill reviewing-release-runbook-bm629
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reviewing-release-runbook
Source: https://github.com/bm629/agent-skills/tree/main/skills/reviewing-release-runbook
Command: npx skills add https://github.com/bm629/agent-skills --skill reviewing-release-runbook-bm629

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Release and deployment runbooks often look complete but fail under pressure: steps an unfamiliar on-call engineer cannot execute, rollbacks that miss a forward change, inlined secrets, or fabricated commands. This Skill acts as an independent acceptance gate that judges a finished runbook against a fixed 11-condition executability and safety bar before it ships. ## Core Features & Use Cases - 11-Condition Review Bar: Judges copy-paste-safe idempotent steps, per-step measurable verification, go/no-go gates, complete strategy-matched rollback, post-deploy verification, escalation and monitoring, no inlined secrets, and no fabricated commands. - Proportional Conditional Checks: Applies delta-scoped amend review, stateful-change safety, and comms-window conditions only when they apply, avoiding false-revises on thin but correct runbooks. - Machine-Parseable Verdict: Emits a single VERDICT: approve or VERDICT: revise line plus actionable findings naming the failed condition, location, and concrete fix. - Use Case: Before approving a production deploy procedure, hand the runbook plus its upstream architecture-doc, technical-design, and test-plan to the reviewer; it spot-checks commands against the upstreams and returns a verdict with fixes the author can resolve in one pass. ## Quick Start Use reviewing-release-runbook to judge the finished runbook in docs/ops/deploy-v2.md against the handed-in architecture-doc and test-plan, and give me the verdict with findings.

Frequently Asked Questions about reviewing-release-runbook

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a release runbook before approving a production deploy?

Hand the finished runbook to the reviewer along with its upstream documents (architecture-doc, technical-design, test-plan). It judges the runbook against 11 executability and safety conditions and emits a VERDICT: approve or VERDICT: revise line with actionable findings.

What does the 11-condition runbook review bar check?

It checks copy-paste-safe idempotent steps, per-step measurable verification, a go/no-go gate, complete strategy-matched rollback with triggers, post-deploy verification reusing test-plan criteria, concrete escalation and monitoring, no inlined secrets, and no fabricated commands, plus conditional amend, stateful-change, and comms checks.

Can this skill review an amend or delta to an existing runbook?

Yes, condition 9 applies a delta-scoped review for changes against an existing runbook. It checks the edit stays in scope, addresses its trigger, does not break downstream preconditions, and includes a version bump and changelog, without re-litigating the unchanged runbook.

Does the runbook reviewer work for small single-service deploys?

Yes, the bar is proportional. A small deploy legitimately collapses sections it does not need, and conditional conditions like stateful-change safety and comms plans are n/a when they do not apply, so a thin but complete runbook is approved rather than false-revised.

When should I not use the release-runbook review gate?

Do not use it to author or fix a runbook, to review a QA test-plan, to gate engineering design documents, or to check CI/CD pipeline configuration. It only judges the operational go-live runbook document, never the live deploy or running service.