A systematic course for developers who want to master Claude Code. Begin with working methods, commands, Memory, and safe recovery; then learn CLI automation, Skills, Hooks, MCP, Subagents, and Plugins; finally combine them into a maintainable, verifiable, reusable Claude Code work system.
Understand Claude Code as a configurable work system and choose the responsibility and learning order of its core capabilities for real scenarios.
Identify high-frequency tasks worth turning into Slash Commands and design reusable commands with arguments, context, permissions, and output constraints.
Place stable project knowledge at the correct Memory layer and write concise, low-noise CLAUDE.md rules that do not conflict across scopes.
Choose appropriate Checkpoint and Rewind strategies for experimentation and correctly separate session recovery from Git version control.
Choose the right CLI mode, permissions, and output format for scripts, pipelines, and CI/CD while controlling unattended-execution risk.
Decide whether work deserves a Skill and design a clear, progressively loaded, permission-bounded, verifiable capability package.
Choose the correct Hook event and handler for automation and design a debuggable safety boundary from inputs, outputs, and exit codes.
Decide whether a need fits MCP and design transport, authentication, scope, tool loading, and troubleshooting boundaries.
Split complex work into Subagent tasks with clear responsibilities, minimal permissions, isolated context, and testable outputs.
Decide when a Plugin is justified and design a complete, secure, installable, testable, distributable team workflow package.
Combine Claude Code capabilities for a real project into a clear, separated, controlled, recoverable, team-maintainable work system.
