deployment-procedures_skill

Overview

This skill teaches production deployment principles and decision-making for safe releases. It focuses on patterns, verification, rollback strategies, and how to think through deployment trade-offs instead of providing one-size-fits-all scripts. The goal is to make deployments predictable and reversible across platforms.

How this skill works

The skill breaks deployments into phases: prepare, backup, deploy, verify, and confirm or rollback. It inspects platform type, verification categories (code quality, build, environment, safety), and recommends platform-appropriate rollback and zero-downtime strategies. It emphasizes decision checkpoints, monitoring windows, and emergency procedures to guide real-time choices during a release.

When to use it

• Before any production release to validate readiness and plan rollback
• When choosing deployment strategy for a new platform or architecture
• During incident response to decide between restart, rollback, or fix-forward
• When designing release processes for microservices, containers, or serverless
• While implementing zero-downtime or canary rollout patterns

Best practices

• Run the 4 verification checks: code quality, build, environment, and safety
• Use small, frequent deploys and feature flags to reduce blast radius
• Document rollback plans and test them before relying on them
• Automate repetitive steps but keep manual checkpoints for high-risk changes
• Notify the team and open monitoring before executing the deployment

Example use cases

• Deploy a web app on a VPS: prepare build, backup data, SSH deploy, verify health endpoint, rollback via restore
• Release a microservice on Kubernetes: push image, kubectl apply, run readiness checks, use kubectl rollout undo if needed
• Perform a risky database migration: backup, run migration in staging, schedule low-traffic window, have rollback script ready
• Introduce a feature via canary: use gradual traffic shifts, monitor errors and latency, roll forward or rollback based on metrics