superpowers-devops

Based on superpowers by Jesse Vincent - the original AI coding agent skill plugin. This repo is a DevOps-focused transformation: same plugin architecture, entirely new skill set built around infrastructure, containers, CI/CD, cloud, networking, code analysis, and debugging.

For kubernetes-skill, adapted from KubeShark by Lukas Niessen - a failure-mode-first, security-hardened Kubernetes agent skill.

A unified DevOps intelligence layer for AI coding agents. Combines AI agent skills that guide you through DevOps workflows with real production-ready tooling templates - all in one plugin.

Install once. Then just work. When you paste an error, a crash log, a Terraform plan, or a broken nginx config, the right skill kicks in automatically and guides you through diagnosis and resolution.

Works in two modes:

• Repo mode - agent has access to your files, scaffolds from templates, modifies configs directly
• Chat mode - no repo needed, paste errors/logs/code and get inline guidance and fix snippets

• Claude Code
• Cursor
• Codex
• OpenCode
• GitHub Copilot CLI
• Gemini CLI

Step 1: Register the marketplace:

/plugin marketplace add tspry/superpowers-devops

Step 2: Install the plugin:

/plugin install superpowers-devops@superpowersdevops

Step 1: Clone the repo:

git clone https://github.com/tspry/superpowers-devops

Step 2: Install from local path:

/plugin install ./superpowers-devops

Search for superpowers-devops in the plugin marketplace, or:

/add-plugin superpowers-devops

Fetch and follow instructions from https://raw.githubusercontent.com/tspry/superpowers-devops/main/.codex/INSTALL.md

Fetch and follow instructions from https://raw.githubusercontent.com/tspry/superpowers-devops/main/.opencode/INSTALL.md

• brainstorming - refine requirements before touching anything
• writing-plans - break work into reviewable tasks
• executing-plans - run plans via subagents
• subagent-driven-development - drive implementation through delegated subagents
• dispatching-parallel-agents - investigate independent problems in parallel
• test-driven-development - write tests before implementation code
• requesting-code-review - submit changes for peer review
• receiving-code-review - process and act on code review feedback
• finishing-a-development-branch - merge and wrap up completed work
• using-git-worktrees - isolated environments for infra changes
• verification-before-completion - confirm changes actually worked
• writing-skills - create or edit agent skill definitions

• infrastructure-provisioning - Terraform, Ansible, Pulumi
• container-operations - Docker
• kubernetes - Kubernetes manifests, Helm, RBAC, NetworkPolicy, security
• ci-cd-pipeline - GitHub Actions, GitLab CI, Jenkins
• cloud-operations - AWS, GCP, Azure
• monitoring-and-alerting - Prometheus, Grafana, logs
• secrets-management - Vault, SOPS, secret scanning
• network-operations - IPs, ports, firewall rules
• reverse-proxy - Nginx, Traefik, HAProxy, SSL
• dns-and-domains - DNS records, propagation debugging
• load-balancing - upstream configs, health checks

• systematic-debugging - root cause before any fix
• code-audit - DevOps readiness + security scan
• code-optimization - memory leaks, loops, performance
• error-reading - stack traces, crash logs, cloud errors

Real templates and configs in tooling/ organized by domain:

tooling/
├── ci/           GitHub Actions, GitLab CI, Jenkins
├── containers/   Dockerfiles, Kubernetes manifests, Helm charts
├── iac/          Terraform, Ansible, Pulumi
├── cloud/        AWS, GCP, Azure
├── monitoring/   Prometheus, Grafana
├── security/     Vault, SOPS, scanning
└── network/      Nginx, Traefik, HAProxy

Each domain has generic templates plus opinionated starters/ for common full-stack setups.

• Run bash tests/validate.sh for a structural check (frontmatter, path references, conventions)
• .pre-commit-config.yaml provides pre-commit hooks for basic validation before commits

MIT