---
name: DevOps Deliverable Checklist
description: Hit every literal requirement a DevOps deliverable/grade checks for.
---
# DevOps Deliverable Checklist

DevOps deliverables (pipeline designs, runbooks, IaC plans, incident postmortems) are graded on completeness of stages, justifications, and concrete commands/configs. Run this checklist before submitting.

## Universal checks
- **Named addressee** if the brief implies one.
- **Word count** inside any band.
- **Labeled sections** using the brief's exact names.

## Pipeline design specific
- List EVERY pipeline stage the brief requires, in order, with rationale for the ordering (fail-fast: cheapest/fastest checks first).
- Specify what each stage does concretely (the linter, the test framework, the build command) — not just "test".
- Compare the requested deployment strategies (blue-green vs canary vs rolling) and RECOMMEND one with justification tied to the service's characteristics.
- State the rollback policy explicitly: every deploy must be reversible, and HOW (the exact mechanism).
- State the secrets-management approach: never in code/files; inject at runtime from a vault/cloud secret store; least privilege.
- State environment promotion: build once, deploy to many; config differs, artifact does not.
- Address monitoring/alerting post-deploy.

## Runbook / incident specific
- Phases in order: detect, mitigate, diagnose, resolve, postmortem.
- Mitigation BEFORE diagnosis.
- Define the incident commander role.
- Postmortem is blameless and produces concrete preventive actions.

## IaC specific
- Declarative, versioned, idempotent, state separated from config.

## Final pass
Re-read the brief; tick each required stage, justification, and concrete config. DevOps rubrics reward specificity ("canary with 1%→10%→100% ramp") over generality ("progressive delivery").