Hands-On Claude Code Tutorial: Project Setup & Essential Commands (Part 1)

Welcome to the first part of our hands-on tutorial series on AI project development with Claude Code! In this series, we’ll use the open-source chinese-poetry-api-main project as a practical example to explore how to leverage Claude Code for efficient coding, project analysis, and environment setup.

Claude Code: Interactive Slash Commands

Claude Code offers a set of interactive slash commands that help control sessions, manage context, and execute tasks directly within a conversation. For example, the /init command initializes Claude Code in your project and generates a CLAUDE.md configuration file—essential for setting up your project environment. Other key commands include /tasks (to list background tasks like sub-agents), /diff (to show uncommitted code changes), and /code-review (to perform multi-dimensional code reviews for errors and improvements).

Claude Code: CLI Startup Commands

Beyond interactive sessions, Claude Code provides CLI startup commands for more control over how sessions are launched. The claude command starts a new interactive session, while claude -c resumes the most recent uncompleted conversation. To restore a specific historical session, use claude -r "session-name" (e.g., claude -r "auth-refactor"). Keeping your tool up-to-date is easy with claude update.

Ant: Official API Client Commands

Ant is the official CLI client for the Claude API, designed for resource operations. Its command structure follows ant [:] [parameters]. For model management, ant models list shows all available Claude models, and ant messages create is the core command for sending messages to Claude (e.g., ant messages create --model claude-opus-4-8 --max-tokens 1024 --message '{role: user, content: "Hello, Claude"}'). Common parameters like --format (for output format) and --debug (for HTTP request/response logs) enhance usability.

Practical Tips for Effective Use

To get the most out of these tools, remember to use the --help flag for any command (e.g., claude --help or ant messages create --help) when in doubt. You can combine parameters—like claude -c -p "check for type errors"—to resume the latest session and run a non-interactive check. Pipe operations are also supported: for example, cat logs.txt | claude -p "explain these logs" lets you analyze log files directly.

Analyzing Projects with Claude Code

The recommended way to analyze a project is to start an interactive Claude Code session in the project’s root directory. Using natural language questions, you can quickly get insights: ask about the tech stack, directory responsibilities, how to run tests, or even find specific code (like user authentication handling).

Sources

• CSDN Blog: peraglobal, "Hands-On AI Project Development with Claude Code (Part 1): Project Environment Initialization" (URL: https://blog.csdn.net/peraglobal/article/details/164184151)