← ClaudeAtlas

containerizationlisted

Containerize applications with Docker and Kubernetes — writing optimized Dockerfiles, multi-stage builds, image layering, and Kubernetes deployment configs. Use when writing or reviewing a Dockerfile, choosing a base image, debugging a container build, or configuring Kubernetes pods and health checks. This skill is specific to Docker containerization and Kubernetes configuration and does NOT cover CI/CD pipelines, monitoring/alerting, or secret management.
ecoma-io/touchstone · ★ 1 · DevOps & Infrastructure · score 63
Install: claude install-skill ecoma-io/touchstone
# Containerization Containers package the application and its dependencies into a reproducible unit. They answer: "will this run the same way everywhere?" ## Image layering Order matters. Least-frequently-changing layers first (OS packages → language runtime → dependencies → application code). A one-line code change should rebuild only the last layer. ## Multi-stage builds Build in one stage, run in another. The production image contains no compilers, no dev dependencies, no source code — only the artifact and its runtime. ## Image size A 2GB image is a deployment delay. Strip package manager caches, remove temp files, use `--no-install-recommends`. The image is the unit of deployment; smaller images deploy faster and have a smaller attack surface. ## Non-root The container must not run as root. `USER 1000` or equivalent. A compromised process running as root owns the node. ## Health checks The container must report its own health. `HEALTHCHECK` in Docker, `livenessProbe`/`readinessProbe` in Kubernetes. A container that is running but not serving is worse than one that crashed — it looks alive but silently fails.