managing-infra — Claude Code skill: what it does, how to install
managing-infra guides Kubernetes architectural choices, Terraform module structuring, Helm usage, and GitHub Actions CI/CD workflows.
MIT 115 installs ★ 17 updated 2026-10-14
npx skills add julianobarbosa/claude-code-skills --skill managing-infra -a claude-codeWhat managing-infra does
This skill supplies your coding assistant with concrete rules for container orchestration and cloud automation. It defines clear boundaries across tools: raw manifests for basic workloads, Kustomize for overlays without templates, and Helm for third-party releases. Workload defaults enforce non-root execution, read-only filesystems, dropped Linux capabilities, and network policies.
A dedicated section covers operational traps across CLI workflows. For example, Kustomize patches that fail to match target resources generate empty diffs silently, demanding explicit verification with kustomize build. Declared tool permissions cover Read, Bash, Grep, and Glob.
What it helps with
- decide whether Helm or Kustomize fits an infrastructure task
- apply non-root security contexts to Kubernetes pods
- structure GitHub Actions pipelines using actions pinned by SHA
- prevent state lock collisions and computed index errors in Terraform
Example prompts for the agent
- Check this Kubernetes deployment for missing security defaults
- Help me structure a Terraform module to avoid state lock contention
- Draft a multi-arch container build workflow for GitHub Actions
How to install managing-infra
1Skills CLI (Vercel)
npx skills add julianobarbosa/claude-code-skills --skill managing-infra -a claude-code2Manually: put the files in place
~/.claude/skills/managing-infra/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/managing-infra/
1Skills CLI (Vercel)
npx skills add julianobarbosa/claude-code-skills --skill managing-infra -a codex2Manually: put the files in place
.agents/skills/managing-infra/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add julianobarbosa/claude-code-skills --skill managing-infra -a cursor2Manually: put the files in place
.cursor/skills/managing-infra/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install julianobarbosa/claude-code-skills skills/managing-infra2Manually: put the files in place
.github/skills/managing-infra/SKILL.md1Skills CLI (Vercel)
npx skills add julianobarbosa/claude-code-skills --skill managing-infra -a gemini-cli2Platform CLI
gemini skills install https://github.com/julianobarbosa/claude-code-skills --path skills/managing-infraManually: put the files in place
.devin/skills/managing-infra/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add julianobarbosa/claude-code-skills --skill managing-infra -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about managing-infra
How to install managing-infra in Claude Code?
Run npx skills add julianobarbosa/claude-code-skills --skill managing-infra -a claude-code. The definition is stored at ~/.claude/skills/managing-infra/SKILL.md.
What is the managing-infra skill used for?
It equips the agent with guidelines for Kubernetes workloads, Terraform modules, build targets, and CI/CD best practices.
Which tool caveats does the skill address?
It details silent empty diffs in Kustomize, schema field drops during Helm upgrades, and conflicts between client-side and server-side kubectl apply.
Source text
SKILL.md · MIT
--- name: managing-infra description: Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices. allowed-tools: Read, Bash, Grep, Glob --- # Infrastructure Patterns ## When to Use What | Tool | Use For | | ------------------ | --------------------------------------------------- | | **Raw K8s YAML** | Simple deployments, one-off resources | | **Kustomize** | Environment variations, overlays without templating | | **Helm** | Complex apps, third-party charts, heavy templating | | **Terraform** | Cloud resources, infrastructure lifecycle | | **GitHub Actions** | CI/CD, automated testing, releases | | **Makefile** | Build automation, self-documenting targets | | **Dockerfile** | Container builds, multi-stage, multi-arch | ## Quick Decisions **Kustomize** when: Simple env differences, readable manifests, patching YAML **Helm** when: Complex templating, third-party charts, release management ## K8s Security Defaults Every workload: non-root user, read-only filesystem, no privilege escalation, dropped capabilities, network policies. ## GitHub Actions Patterns - **CI workflow**: Lint, test, compile on PRs (run on both x86 + ARM) - **Release workflow**: Multi-arch Docker build on tags (native ARM runners) - Pin actions by SHA, least-privilege permissions ## References - [KUBERNETES.md](KUBERNETES.md) - K8s resource patterns - [TERRAFORM.md](TERRAFORM.md) - Terraform module patterns - [GITHUB-ACTIONS.md](GITHUB-ACTIONS.md) - CI/CD workflow patterns - [MAKEFILE.md](MAKEFILE.md) - Build automation patterns - [DOCKERFILE.md](DOCKERFILE.md) - Container build patterns - [templates/](templates/) - Ready-to-use templates ## Commands ```bash kubectl apply -k ./ # Apply kustomize helm upgrade --install NAME . # Install/upgrade chart terraform plan && terraform apply ``` --- ## Gotchas - **Terraform state lock contention**: default 10-min lock timeout; bumped timeout doesn't help if the lock holder hung — force-unlock only after confirming the process is dead. - **Helm release name reuse on uninstalled-but-not-purged release** fails install with "already exists" — use `--no-hooks` + explicit purge, or never reuse names. - **Kustomize patches that match nothing silently produce empty diffs** — verify with `kustomize build` after every patch addition. - **Terraform `for_each` over a computed value forces apply-time count** — can cause spurious re-creation of resources between plans. - **`helm upgrade --install` on a changed values schema** can silently drop fields that no longer match — diff the rendered output, not just the values file. - **`kubectl apply --server-side` vs client-side conflicts** when both have been used: client-side last-applied-config can shadow server-side managed fields without error.
© julianobarbosa · MIT · julianobarbosa/claude-code-skills