AI CODING AGENT / TERMINAL WORKFLOW / 06 DEEP DIVES
Claude Code principles and practice
Understand Claude Code as a terminal-first coding agent that reads projects, runs commands, edits files, and verifies results; then apply memory, permission modes, sessions, hooks, MCP, subagents, and team adoption practices. The guide body is currently maintained in Chinese.
The reviewed guide body for this track is currently maintained in Chinese.Curriculum
6 deep dives
- 00Claude Code principles and architectureA Chinese guide to Claude Code's context-action-verification loop across terminal projects and model reasoning.
- 01Installation and project memorySet up Claude Code with installation, authentication, working directories, CLAUDE.md, auto memory, and context inspection.
- 02Daily development workflowUse Claude Code for daily development through problem framing, exploration, planning, edits, verification, session resume, and delivery notes.
- 03Codebase reading and refactoringRead large codebases, locate behavior boundaries, split refactors, and keep changes isolated with Claude Code.
- 04Hooks, MCP, subagents, and automationApply Claude Code hooks, MCP, subagents, skills, plugins, and non-interactive runs to real workflows.
- 05Team adoption, security, and costAdopt Claude Code in teams with permission modes, protected paths, data retention, collaboration, cost control, and CI/CD boundaries.