Production-ready Claude Code and Codex skills for DevOps, SRE, and platform engineering teams.
| Skill | Description | Cloud / Stack |
|---|---|---|
| zombie-resource-scanner | Scan cloud infrastructure for zombie/idle resources, generate cost optimization recommendations with auto-classification and scheduled task protection | AWS, Alibaba Cloud, Azure (WIP) |
Existing DevOps skill projects focus on coding workflows — Terraform authoring, CI/CD config, K8s YAML. This project covers operational workflows: what SRE teams actually do day-to-day — scanning zombie resources, troubleshooting incidents, analyzing blast radius, reviewing monitoring coverage.
Each skill follows a multi-agent pipeline pattern:
Plan → Collect → Validate → Analyze → Report
Rather than a single flat prompt, each skill orchestrates specialized agents that verify each other's work, producing reliable, production-safe results.
# Claude Code
cp -r skills/* ~/.claude/skills/
# Codex
cp -r skills/* ~/.codex/skills/Or cherry-pick individual skill directories.
- K8s troubleshooting skill (pod failures, cluster issues, OOM diagnosis)
- Monitoring coverage audit (Prometheus/Grafana gap analysis)
- Incident blast-radius calculator
- CI/CD pipeline health check
- AWS cost optimization (Reserved Instance / Savings Plan analysis)
- Secret sprawl detection
Skills use standard Claude Code SKILL.md format. Each skill directory is self-contained with zero external dependencies beyond the agent runtime.
- Fork the repo
- Create
skills/your-skill-name/SKILL.md - Add tests and examples under
skills/your-skill-name/ - Open a PR with a description of what the skill does and who it's for
MIT