This project offers comprehensive reverse-engineering documentation for Claude Code CLI Node.js v2.1.84. It serves as a knowledge base for understanding the CLI's core runtime logic, module boundaries, and ecosystem, facilitating learning, analysis, and rewrite efforts without providing original source code or a runnable alternative.

How It Works

HitCC is a documentation knowledge base derived from static analysis of the Claude Code CLI v2.1.84 package. It meticulously details the core runtime logic, module boundaries, configuration, and ecosystem. The content is organized by runtime topics, not the original source tree, aiming to provide a stable reference for building runnable alternatives or high-similarity rewrites.

Quick Start & Requirements

Obtain the analyzed CLI package via npm pack @anthropic-ai/claude-code@2.1.84. Python scripts in recovery_tools/ aid in cleaning obfuscated code. This repository relies solely on static analysis; dynamic analysis or network service usage was not employed.

Highlighted Details

• Structured reverse-engineering documentation covers the CLI's execution chain: startup, agent loop, tool use, prompt assembly, and session persistence.
• Topic-oriented analysis details surrounding systems (MCP, Plugin, Skill, TUI, Remote Persistence, Bridge), including candidate architectures and open questions for rewrites.
• Documentation spans 81 files (27,170 lines, 698,707 characters) in docs/, organized by runtime topics.
• Sufficiently detailed to support reconstructing a runnable alternative or a modular, high-similarity rewrite.

Maintenance & Community

Maintained by "Hitmux contributors" with support from the "Linux Do community." No specific community links or roadmap details are provided.

Licensing & Compatibility

Licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). Allows copying, redistribution, modification, and commercial/non-commercial use, provided attribution, license link, and change indication are included.

Limitations & Caveats

This repository does not contain original source code, a runnable CLI/SDK, or installation scripts. It does not reconstruct the original file tree, private server-side logic, or guarantee 1:1 edge-case reproduction. It is based exclusively on static analysis.