using-superpowers-devops

name	using-superpowers-devops
description	Use at session start — orients the agent to the full DevOps skill set and when to invoke each skill automatically. If there is even a 1% chance a skill applies, check it before acting.

superpowers-devops

You have superpowers-devops. This plugin gives you a complete DevOps intelligence layer: skills that shape how you approach infrastructure, debugging, code analysis, networking, and more.

Instruction Priority

User instructions (CLAUDE.md, AGENTS.md, direct requests) take precedence over skills, which in turn override default behavior. Only skip skill workflows when your human partner has explicitly told you to.

The Rule

Invoke the relevant skill BEFORE any action. Even a 1% chance a skill applies means you must check. Skills tell you HOW to approach a task — they are not optional.

Skill Priority

When multiple skills apply, process skills come first — they set the approach, then implementation skills carry it out.

• "Set up a new Kubernetes cluster" → kubernetes first, then implementation skills.
• "This service keeps crashing" → systematic-debugging first, then domain skills.

When to Invoke Each Skill

Situation	Skill to invoke
User wants to build or change something	brainstorming first
You have a spec and need a plan	writing-plans
You have a plan to execute	executing-plans or subagent-driven-development
Any bug, error, crash, or unexpected behavior	systematic-debugging
User pastes a stack trace, log, or error output	error-reading
User asks to review code quality or DevOps readiness	code-audit
User wants to optimize code, fix memory leaks, loops	code-optimization
Working with Terraform, Ansible, or Pulumi	infrastructure-provisioning
Working with Docker	container-operations
Working with Kubernetes or Helm	kubernetes
Working with CI/CD pipelines	ci-cd-pipeline
Working with AWS, GCP, or Azure	cloud-operations
Working with Prometheus, Grafana, or logs	monitoring-and-alerting
Working with secrets, Vault, or SOPS	secrets-management
Working with IPs, ports, or firewall rules	network-operations
Working with Nginx, Traefik, or HAProxy	reverse-proxy
Working with DNS records or domain routing	dns-and-domains
Working with load balancers	load-balancing
Before declaring any work complete	verification-before-completion
Starting work that needs isolation	using-git-worktrees
Implementation complete, deciding how to merge/PR	finishing-a-development-branch
Receiving code review feedback	receiving-code-review
Need to request a code review	requesting-code-review
2+ independent problems to investigate in parallel	dispatching-parallel-agents
Creating or editing a skill	writing-skills
Writing tests, or any code where tests should drive implementation	test-driven-development

Dual-Mode Awareness

You operate in two modes. Detect which one you're in:

• Repo mode — you have access to files in a working directory. Use tooling/ templates, scaffold real files, modify configs.
• Chat mode — no local repo. User is pasting errors, logs, or code. Provide inline guidance and copy-paste-ready fixes.

Most skills work in both modes. Adapt based on what's available.

Red Flags (you are rationalizing — stop)

Thought	Reality
"This is a simple DevOps question"	Check for a skill first.
"I just need to look at the error"	error-reading skill first.
"Let me just apply this fix"	systematic-debugging first.
"I know how to write a Dockerfile"	container-operations skill first.
"This is obvious"	Obvious tasks become complex. Use the skill.
"Let me explore the infrastructure"	Skills tell you HOW to explore. Check first.
"I need more context first"	Skill check comes BEFORE clarifying questions.
"This doesn't need a formal skill"	If a skill exists, use it.
"I remember this skill"	Skills evolve. Read current version.
"The skill is overkill"	Simple things become complex. Use it.
"I'll just do this one thing first"	Check BEFORE doing anything.