A Chinese field manual for terminal-shy AI coders

What it does

This is a curated Chinese-language tutorial collection—92 articles split between Anthropic’s Claude Code and OpenAI’s Codex CLI. It walks beginners from basic command-line concepts through agent loops, MCP, sandbox approvals, and migration strategies. The author claims every feature is cross-checked against official documentation rather than third-party guesswork.

The interesting bit

The guide refuses to be a dry translation. Each article includes first-person pitfall stories with concrete details and real numbers, and teaches via a three-part structure: scenario setup, everyday analogy, then a concrete command-line task. Even the 64 custom SVG illustrations use a dark Catppuccin Mocha theme so the site mimics the actual CLI window.

Key highlights

• 53 Claude Code articles and 39 Codex articles, covering agent loops, MCP, Skills, Hooks, sandbox approvals, and GitHub Actions integration.
• A dedicated “basics” track for command-line novices; the FAQ explicitly says no prior terminal experience is required.
• Content is free under MIT, though the tools themselves require a paid subscription or API access starting around $20 per month.
• Requires proxy/VPN access for installation and login; the guide flags this upfront rather than letting users hit a wall mid-setup.
• Packaged as a VitePress site with Shiki syntax highlighting, mermaid diagrams, and auto-generated OG cards.

Caveats

• The entire repository and tutorial site are written in Chinese, which is great if you read the language and useless otherwise.
• Both target tools iterate rapidly, so the guide’s accuracy decays unless the author keeps patching; the commit log doubles as the changelog.

Verdict

Grab this if you are a Chinese-speaking developer who wants to use Claude Code or Codex but needs a gentle, verified on-ramp. Skip it if you already live in a terminal and prefer to read English official docs directly.