claude4ops — DevOps Superpowers for Everyone

Production-ready DevOps superpowers for everyone. Guided workflows for Kubernetes, Terraform, CI/CD, SLOs, and incident response — with token optimization built in. Runs on Linux, macOS, and Windows.

Why claude4ops?

DevOps workflows are repetitive, error-prone, and hard to get right. claude4ops brings structured, production-tested procedures directly into Claude Code so you skip the RTFM loop and ship safely:

• Kubernetes — deploy, validate manifests, and debug pods without memorizing kubectl flags
• Terraform — review plans for risk and cost impact before apply, with state backup and rollback
• CI/CD — generate GitHub Actions / GitLab pipelines, design blue/green or canary strategies, ship features end-to-end
• Observability — define SLOs with error budgets, create SLO-based alerts with burn-rate thresholds and runbook links
• Incident response — run blameless postmortems with structured timelines, RCA, and action items
• Multi-cloud — all commands cover AWS, GCP, and Azure with provider-specific examples
• Multi-platform — native support for Linux, macOS, and Windows (PowerShell, WSL2, or Git Bash)

Installation

Quick Install (Recommended)

/plugin marketplace add ops4life/claude4ops
/plugin install claude4ops

From Local Source

git clone https://github.com/ops4life/claude4ops.git
# In Claude Code:
# /plugin marketplace add /absolute/path/to/claude4ops
# /plugin install claude4ops

Setup

After installing the plugin, run /claude4ops:install to configure Claude Code:

• Scope: user (~/.claude/) or project (.claude/)
• Settings: settings.json merge + status line script
• Hooks: block-prod, auto-lint, audit, Slack notifications
• MCP servers: AWS, Kubernetes, GitHub
• Skills: docling (PDF/DOCX/image → Markdown), convert (Claude Code → Gemini/Copilot/Cursor/Codex)
• Rules: git workflow standards
• Optimization: token-efficient tools — RTK shell proxy (60-90% savings) + Caveman compressed mode (~75% savings)
• Plugins: context7, playwright, superpowers, frontend-design

Architecture & Components

Platform Support

Platform	Shell	Hook scripts
Linux	bash / sh	.sh
macOS	bash / sh	.sh
Windows	PowerShell (pwsh)	.ps1 — invoked via pwsh -NoProfile -File
Windows	WSL2 or Git Bash	.sh

Prerequisites: jq — apt-get install jq · brew install jq · winget install stedolan.jq. Optional on Windows PowerShell (falls back to built-in JSON).

Windows one-time setup: allow script execution before running /claude4ops:install:

Set-ExecutionPolicy -Scope CurrentUser RemoteSigned

Commands

All commands: /claude4ops:<category>:<command>

Kubernetes

Command	Description
/claude4ops:k8s:deploy	Guided deployment with pre/post validation and rollback
/claude4ops:k8s:troubleshoot	Systematic debugging for pods, services, and network issues
/claude4ops:k8s:manifest-validate	YAML validation for syntax, security, and best practices

Terraform

Command	Description
/claude4ops:terraform:plan-review	Risk, security, and cost analysis of a Terraform plan
/claude4ops:terraform:apply	Safe apply with state backup and rollback procedures
/claude4ops:terraform:cloud-cost	Multi-cloud cost analysis and right-sizing recommendations

CI/CD

Command	Description
/claude4ops:cicd:pipeline-new	Generate production CI/CD pipeline (GitHub Actions, GitLab, Jenkins)
/claude4ops:cicd:ship	Full agentic pipeline: lint → test → confirm → commit → push
/claude4ops:cicd:pr-review	DevOps-focused review: Terraform, secrets, containers, pipelines
/claude4ops:cicd:deploy-strategy	Design blue/green, canary, or rolling deployment strategy

Observability

Command	Description
/claude4ops:observability:slo-define	Define SLOs/SLIs with error budgets and burn-rate alerting
/claude4ops:observability:alert-new	Create SLO-based monitoring alerts with runbook links

Install & Automation

Command	Description
/claude4ops:install	Install hooks, skills, rules, settings, plugins, and token optimization tools — user or project scope
/claude4ops:hooks:mcp-setup	Configure MCP servers for AWS, Kubernetes, and GitHub

Incident Management

Command	Description
/claude4ops:incident:postmortem	Blameless postmortem with timeline, RCA, and action items

Common Workflows

Deploy to Kubernetes:

/claude4ops:k8s:manifest-validate → /claude4ops:k8s:deploy → /claude4ops:k8s:troubleshoot

Infrastructure change:

/claude4ops:terraform:plan-review → /claude4ops:terraform:apply → /claude4ops:terraform:cloud-cost

Feature ship:

/claude4ops:cicd:pr-review main → work on feature → /claude4ops:cicd:ship feat(api): add rate limiting

First-time setup (do once):

/claude4ops:install  # select All to install everything, or pick components individually
/claude4ops:hooks:mcp-setup

Contributing

See CONTRIBUTING.md for the full guide: setup, command structure, worked example, commit format, and PR process.

Resources

• CONTRIBUTING.md — contributor guide
• CLAUDE.md — plugin architecture
• .github/RELEASING.md — release process
• Edmund's Claude Code — reference implementation

License

MIT — see LICENSE