learn-claude-code

This repository stores a local research copy of @anthropic-ai/claude-code@2.1.88, including:

1. npm-original/

   • Original npm tarball and key extracted package artifacts
   • cli.js.map used for source reconstruction

2. source/

   • Reconstructed source tree from cli.js.map
   • Focused on Claude Code's own source (third-party node_modules removed)

3. doc/

   • Analysis reports, module mapping, structure summary, and topic-focused notes

Language

• English: README.md
• 中文：README.zh-CN.md

Layout

• npm-original/ — raw package artifacts
• source/ — reconstructed source code
• doc/ — analysis documents
• README.md — English overview
• README.zh-CN.md — 中文说明

Module analysis index

• Authentication & Login

  • English: doc/authentication-login.en.md
  • 中文：doc/authentication-login.zh-CN.md

• Permissions & Risk Control

  • English: doc/permissions-risk-control.en.md
  • 中文：doc/permissions-risk-control.zh-CN.md

More module reports will be added incrementally: multi-agent, MCP, remote/bridge, telemetry, and update/install.

Notes

• Package: @anthropic-ai/claude-code
• Version: 2.1.88
• Source reconstruction is based on sources + sourcesContent embedded in cli.js.map
• This repo is organized for inspection and learning
• This is a research and analysis repository, not an official source mirror

Current high-level impression

From the recovered source tree, Claude Code 2.1.88 appears to be much more than a thin terminal wrapper. It includes:

• a full OAuth / API-key authentication stack
• permission approval and risk-control logic
• multi-agent / teammate / swarm collaboration features
• deep MCP integration
• remote-control / bridge / direct-connect capabilities
• telemetry / analytics / experiment infrastructure
• update and local installation logic