Mastering Claude: From Conversation to Claude Code in 5 Days

Taught in Chinese

Week 1 · From Conversation to Claude Code

Advanced Prompting and Claude's "Personality": System Prompt, XML Tags, Letting the Model Think First, Structured Output

Skip the prompting basics and go straight to four Claude-specific habits: using a system prompt to set the role, using XML tags to organize long inputs, letting the model think before it answers, and requesting structured output.

Long Documents, Multimodal Input, and a First Look at the API: Using Large Context, Saving Money With Prompt Caching, PDF and Image Input, Citation-Backed Answers; a Minimal Messages API Call

Make your first code call to Claude: feed it a whole PDF and get back a summary with page-numbered citations, then use prompt caching to cut the cost of repeated questions by an order of magnitude.

Getting Started With Claude Code and Managing Context: Install, Writing CLAUDE.md and "Trim Until You Can't", Permission Modes, Plan Mode's "Explore, Then Plan, Then Write", /clear /compact /rewind, Giving Claude a Verifiable Check

Install Claude Code, write the first CLAUDE.md for your own project, learn Plan Mode's explore-then-plan flow, manage context with /clear /compact /rewind, and give Claude a check it can run on its own.

Extending Claude Code: Hooks (Deterministic) vs. CLAUDE.md (Advisory), Skills, Subagents, Plugins, Wiring Up an MCP Server, CLI Tools First

CLAUDE.md is only advisory; hooks are the real gate. Learn to make "must happen every time" rules deterministic with hooks, package reusable workflows as a skill, delegate research to a subagent, then wire up MCP and CLI tools.

Automation and Scale: Headless -p Into CI, Parallel Sessions and Worktrees, Writer/Reviewer Dual Sessions, Adversarial Review, Common Failure Modes; a 20-Line Minimal Agent SDK Agent

Turn Claude Code from a terminal sidekick into one stage of a pipeline: run it headless with -p in CI and parse the JSON result, run parallel sessions with worktrees, use adversarial review as a backstop, then write a 20-line minimal agent with the Agent SDK.