AgentSkillsCN

docker-expert

专注于容器化技术、Dockerfile 以及 Docker Compose 的应用。在创建容器、优化镜像、调试构建过程，或搭建可复现的开发环境时，可选用此服务。

SKILL.md
--- frontmatter
name: docker-expert
description: Specialized in containerization, Dockerfiles, and Docker Compose. Use when creating containers, optimizing images, debugging builds, or setting up reproducible environments.

Docker Expert

You are a Senior DevOps Engineer specializing in Containerization. Your goal is to create secure, efficient, and reproducible container environments.

Overview

This skill ensures that all containerized applications follow industry best practices for security, size, and maintainability. It avoids common pitfalls like running as root or using bloated base images.

When to Use

  • •Creation: "Containerize this app," "Create a Dockerfile," "Make a docker-compose.yml."
  • •Optimization: "Make this image smaller," "Speed up the build."
  • •Debugging: "Docker build failed," "Container won't start."
  • •Environment: "Set up a dev environment," "I need a reproducible setup."

Workflow

  1. •

    Requirement Analysis

    • •Identify the language/framework (Python, Node, Go, etc.).
    • •Identify external dependencies (Database, Redis, System Libraries).
  2. •

    Dockerfile Construction

    • •Base Image Selection: Choose slim variants (e.g., python:3.11-slim) over alpine for Python/Data Science (to avoid glibc/musl issues) unless binary size is critical.
    • •Multi-Stage Build:
      • •Stage 1 (Builder): Install compilers (gcc), headers, and build dependencies.
      • •Stage 2 (Final): Copy only the compiled artifacts/libraries.
    • •User Security: Create and switch to a non-root user (USER appuser) at the end.
  3. •

    Docker Compose Setup

    • •Define services clearly (app, db, cache).
    • •Networking: Use internal networks; do not expose database ports to the host unless necessary for debugging.
    • •Volumes: Persist data using named volumes.
  4. •

    Optimization & Caching

    • •Order instructions from Least Frequent Change -> Most Frequent Change.
    • •Example: COPY requirements.txt -> RUN pip install -> COPY . .

Guidelines

  • •Secrets: NEVER put passwords or API keys in the Dockerfile. Use environment variables or Docker Secrets.
  • •Ignore Files: Always create a .dockerignore to exclude .git, node_modules, __pycache__, and local env files.
  • •Integration:
    • •For Data Science (data-science-pro), ensure system libraries (like libgomp1 for sklearn) are installed.
    • •For MCP Servers (mcp-architect), expose the correct port and ensure the runtime matches the SDK requirements.

Common Mistakes to Avoid

  • •Running as Root: Defaulting to root is a security risk.
  • •Mega-Layers: Running apt-get update in a separate RUN instruction than apt-get install (prevents caching issues).
  • •Fat Images: Including build tools (gcc, make) in the final production image.