Skillget

open-claude-code

skill
Guvenlik Denetimi
Basarisiz
Health Gecti
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 75 GitHub stars
Code Basarisiz
  • process.env — Environment variable access in open_claude_code/src/api/client.mjs
  • network request — Outbound network request in open_claude_code/src/api/client.mjs
  • process.env — Environment variable access in open_claude_code/src/config/env.mjs
  • os.homedir — User home directory access in open_claude_code/src/config/paths.mjs
Permissions Gecti
  • Permissions — No dangerous permissions requested
Purpose
This tool is a clean-room, open-source implementation of Anthropic's Claude Code CLI, rebuilt from decompilation analysis. It provides a terminal-based interface for interacting with the Claude API, featuring conversation management, file editing, and MCP client capabilities.

Security Assessment
The tool accesses the user's home directory (FAIL) to establish configuration paths and reads environment variables (WARN) to retrieve necessary API keys for authentication. It makes outbound network requests (WARN) to communicate with the Claude API. There are no hardcoded secrets and no dangerous system permissions are requested. However, because it is built using reverse-engineered intelligence and features system-level tools (such as Bash execution and custom agents), it inherently interacts closely with your local environment. Overall risk is rated as Medium.

Quality Assessment
The project is highly active, with its last push occurring today. It is protected by the permissive MIT license and has garnered 75 GitHub stars, indicating a moderate level of community trust and interest.

Verdict
Use with caution: while the project is active, properly licensed, and free of hardcoded secrets, its reverse-engineered origins and system-level file access require users to thoroughly inspect the code before integrating it into sensitive workflows.
SUMMARY

Open Source Claude Code CLI — Reverse Engineered & Rebuilt

README.md

Open Claude Code

Open Source Claude Code CLI — Reverse Engineered & Rebuilt

The open source implementation of Anthropic's Claude Code CLI,
built from decompiled source intelligence using ruDevolution.

Node.js License Status Based on

🔥 Background: The Claude Code Source Leak

On March 31, 2026, Anthropic accidentally shipped source maps in the Claude Code npm package, exposing the full TypeScript source. The leak revealed:

  • KAIROS — an autonomous agent system that works without user input
  • Undercover Mode — hides AI involvement in commits from Anthropic employees
  • Internal tools (TungstenTool, SendUserFileTool, PushNotificationTool)
  • 22 private repository names
  • Animal codenames for unreleased models

Anthropic patched it within hours, but the architecture was documented by Sabrina Ramonov and others.

This project takes a different approach. Instead of relying on leaked source, we use ruDevolution — an AI-powered decompiler — to analyze the published npm package legally, and rebuild an open source version from that intelligence.

📦 What Is This?

Open Claude Code is a clean-room open source implementation of the Claude Code CLI architecture. It's not a copy of Anthropic's code — it's a ground-up rebuild informed by decompilation analysis of the published binary.

v1 (Current — 2025)

The original implementation with basic WASM terminal UI, conversation management, and Claude API integration. Built before the decompilation intelligence was available.

v2 (Coming Soon — 2026)

A complete rewrite based on ruDevolution's decompilation of Claude Code v2.1.91 (34,759 declarations, 981 modules). The v2 architecture mirrors the actual Claude Code internals:

  • Async generator agent loop — 13 event types, recursive after tool execution
  • 25+ built-in tools — Bash, Read, Edit, Write, Glob, Grep, Agent, WebFetch
  • 6 permission modes — bypassPermissions, acceptEdits, auto, default, dontAsk, plan
  • MCP client — stdio, SSE, Streamable HTTP transports
  • Hooks system — PreToolUse, PostToolUse, Stop events
  • Settings chain — user/project/local/managed
  • Context compaction — automatic context window management
  • Custom agents & skills — extensible AI personas

Preview the v2 source →

🔍 What ruDevolution Found in Claude Code

Our decompiler discovered capabilities Anthropic hasn't publicly announced:

Discovery Evidence
🤖 Agent Teams CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS, TEAMMATE_COMMAND
🌙 Auto Dream Mode tengu_auto_dream_completed — works while you sleep
🔮 claude-opus-4-6 Unreleased model ID (current public is 4.5)
🔮 claude-sonnet-4-6 Unreleased model ID
🔐 6 "amber" codenames amber_flint, amber_prism, amber_stoat, amber_wren
🧰 Advisor Tool tengu_advisor_tool_call — new tool type
🧰 Agentic Search Search that spawns sub-agents
☁️ CCR (Cloud Code Runner) Full cloud execution with BYOC
🎮 Powerups Gamification with unlockable abilities
📡 MCP Streamable HTTP New transport replacing SSE
📱 Chrome Extension Extension ID: fcoeoabgfenejglbffodgkkbkcdhcgfn
🎙️ Native Audio Voice input capability
117 new env vars Since v2.0

Download decompiled releases →

⚡ Quick Start

v1 (Current)

git clone https://github.com/ruvnet/open-claude-code.git
cd open-claude-code/open_claude_code/src
npm install
ANTHROPIC_API_KEY=your-key node index.mjs

v2 (Preview)

cd open-claude-code/v2
ANTHROPIC_API_KEY=your-key node src/index.mjs "explain this codebase"

Decompile Claude Code Yourself

npx ruvector decompile @anthropic-ai/claude-code

🏗️ v2 Architecture

v2/src/
├── core/
│   └── agent-loop.mjs        # Async generator (13 event types)
├── tools/
│   ├── registry.mjs           # validateInput/call interface
│   ├── bash.mjs, read.mjs     # Built-in tools
│   ├── edit.mjs, write.mjs
│   └── glob.mjs, grep.mjs
├── permissions/
│   └── checker.mjs            # 6 permission modes
├── config/
│   ├── settings.mjs           # User/project/local chain
│   └── cli-args.mjs
└── index.mjs                  # Entry point

⚖️ Legal

This project is a clean-room implementation — not a copy of Anthropic's source code. The architecture is informed by analysis of the published npm package using ruDevolution, which is legal under:

  • 🇺🇸 US DMCA §1201(f) — reverse engineering for interoperability
  • 🇪🇺 EU Software Directive Art. 6 — decompilation for interoperability
  • 🇬🇧 UK CDPA §50B — decompilation for interoperability

No leaked source code was used. No DRM was bypassed. No proprietary code was copied.

🔗 Related

📄 License

MIT

Yorumlar (0)

Sonuc bulunamadi