1. Conduid
  2. AI
  3. Claude Code Leak
MCP server · AI

Claude Code Leak

About Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands. Today it was leaked, so we are using this repo to work with it

Unclaimed ai
34Low

Scored a month ago · breakdown

About Claude Code Leak

Claude Code Leak is an MCP server in the AI category: about Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands. Today it was leaked, so we are using this repo to work with it.

Install

From source
git clone https://github.com/soloxiao/claude-code-leak

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Claude Code Leak

Powered by Claude · Grounded in docs

I know everything about Claude Code Leak. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

  • ✓README presentThe repository ships documentation.

5 checks have not been run yet.

README

Claude Code Source Repository Summary

中文

Overview

This repository contains the complete client-side source code of Claude Code — Anthropic's official AI-powered coding assistant CLI tool. It provides a terminal-based interactive interface for developers to collaborate with Claude AI on reading, writing, editing, and analyzing code through conversation.

Codebase Scale

Metric Value
Total source files 1,904
TypeScript/TSX files 1,884
Total lines of code ~512,664
Source directory size 33 MB
Top-level module directories 36

Tech Stack

Category Technology
Runtime Bun
UI Framework React + Ink (terminal renderer)
CLI Framework Commander.js
API Client @anthropic-ai/sdk
Schema Validation Zod v4
MCP Integration @modelcontextprotocol/sdk
Feature Flags GrowthBook + Bun compile-time feature()
State Management Custom React Context Store (useSyncExternalStore)

Ten Core Modules

1. Interactive REPL

Entry chain: cli.tsx → main.tsx (4,683 lines) → replLauncher.tsx. Handles CLI argument parsing, session initialization, authentication, and REPL startup. ~146 UI component files.

2. Conversation & Query Engine

query.ts (1,729 lines) + QueryEngine.ts (1,295 lines). Manages message construction, API calls, tool execution, automatic context compression, and token/cost tracking.

3. Tool System (43 Tools)

Located under src/tools/:

Category Tools
Shell Execution BashTool (with AST security analysis & sandbox), PowerShellTool
File Operations FileReadTool, FileWriteTool, FileEditTool
Search GlobTool, GrepTool
LSP Integration LSPTool (go-to-definition, find references, hover info, etc.)
Web Interaction WebFetchTool, WebSearchTool
Agent System AgentTool (subagent spawning: local/remote/fork/worktree)
Task Management TaskCreateTool, TaskGetTool, TaskListTool, TaskUpdateTool, TaskStopTool, TaskOutputTool
Team Collaboration TeamCreateTool, TeamDeleteTool, SendMessageTool
Plan Mode EnterPlanModeTool, ExitPlanModeTool
Worktree EnterWorktreeTool, ExitWorktreeTool
MCP MCPTool, ListMcpResourcesTool, ReadMcpResourceTool
Scheduled Tasks ScheduleCronTool (CronCreate/Delete/List)
Remote Triggers RemoteTriggerTool
Notebook NotebookEditTool
User Interaction AskUserQuestionTool
Other SkillTool, ConfigTool, SleepTool, BriefTool

4. Slash Command System (100+ Commands)

Located in src/commands/:

Category Example Commands
Session Management /resume, /session, /compact, /clear
Git Integration /commit, /commit-push-pr, /branch, /diff
Configuration /config, /permissions, /theme, /color, /effort
MCP Management /mcp
Plugins /plugin, /reload-plugins
Authentication /login, /logout, /install-github-app, /install-slack-app
Code Review /review, /security-review
Remote Collaboration /bridge, /teleport, /remote-setup
Advanced Features /ultraplan, /thinkback, /plan, /tasks, /vim, /voice
Utilities /doctor, /help, /status, /stats, /usage, /cost, /export, /copy

5. Multi-Agent System

"Swarm" architecture supporting subagents working in parallel within isolated Git worktrees, with team creation/deletion/messaging. The src/coordinator/ module handles multi-agent orchestration.

6. Remote Sessions (Claude Code Remote)

src/bridge/ (33 files). WebSocket + JWT authentication based, supporting session management and persistence in remote containers.

7. Permission System

Fine-grained tool permission control, Bash command AST-level security analysis (parseForSecurity), policy limits and rate control, permission denial tracking.

8. Context Management

Automatic aggregation of Git context, CLAUDE.md project configuration, and persistent Memory system. src/services/compact/ provides automatic conversation compression.

9. Plugin & Skill System

Extensible architecture: hot-reloading plugins (src/plugins/), built-in + MCP-extended skills (src/skills/), MCP server connection management (src/services/mcp/).

10. Voice Input

Speech-to-text integration (src/voice/, src/services/voice.ts), controlled via feature('VOICE_MODE') feature flag.


Directory Structure

src/
├── entrypoints/       # Application entry points
├── main.tsx           # Main entry (4,683 lines)
├── query.ts           # Core conversation loop
├── QueryEngine.ts     # Query orchestration engine
├── Tool.ts            # Tool type system
├── tools/             # 43 tool implementations
├── commands/          # 100+ slash commands
├── components/        # 146 UI components
├── hooks/             # 85+ React hooks
├── services/          # Backend services (API, MCP, OAuth, analytics, etc.)
├── utils/             # 330+ utility functions
├── state/             # State management
├── bridge/            # Remote bridging
├── types/             # TypeScript type definitions
├── coordinator/       # Multi-agent coordinator
├── skills/            # Skill system
├── plugins/           # Plugin system
├── screens/           # Top-level page components
├── server/            # Remote session server
├── remote/            # Direct connection features
├── context/           # React Context
├── constants/         # Constants
├── migrations/        # Data migrations
├── keybindings/       # Keybinding system
├── vim/               # Vim mode
├── voice/             # Voice input
├── ink/               # Ink extension utilities
├── memdir/            # Memory directory management
└── buddy/             # Buddy sprite (animated character)

Feature Flags

Flag Description
COORDINATOR_MODE Multi-agent coordinator mode
VOICE_MODE Voice input
BRIDGE_MODE Remote bridge mode
PROACTIVE Proactive suggestions
AGENT_TRIGGERS Agent triggers
MONITOR_TOOL Monitor tool
KAIROS Kairos feature

Security Mechanisms

  • Bash command AST parsing & security analysis
  • Sandbox execution adapters
  • Read-only validation
  • Fine-grained permission control & denial tracking
  • Anthropic internal features isolated via process.env.USER_TYPE === 'ant'

Service Layer Details (src/services/)

Module Description
api/ Claude API client wrapper
oauth/ OAuth authentication flow
mcp/ MCP server connection & management
lsp/ Language Server Protocol integration
compact/ Automatic conversation compression
analytics/ Usage analytics & tracking
plugins/ Plugin services
SessionMemory/ Session memory persistence
teamMemorySync/ Team memory synchronization
extractMemories/ Automatic memory extraction
MagicDocs/ Intelligent documentation service
PromptSuggestion/ Prompt suggestions
AgentSummary/ Agent summaries
tools/ Tool service layer
tokenEstimation.ts Token estimation
policyLimits/ Policy & limit management
remoteManagedSettings/ Remote configuration management
settingsSync/ Settings synchronization
tips/ Tips & suggestions
voice.ts Voice service
vcr.ts Session recording/playback

Additional Modules

Module Path Description
Buddy Sprite src/buddy/ Animated terminal companion character
Vim Mode src/vim/ Full Vim keybindings (motions/operators/text objects)
Output Styles src/outputStyles/ Switchable output display modes
Upstream Proxy src/upstreamproxy/ HTTP proxy support
Native TS src/native-ts/ Native TypeScript utilities
Moreright src/moreright/ Extended permission module
Keybindings src/keybindings/ Customizable keybinding system
Migrations src/migrations/ Configuration/data version migrations

Platform Support

Claude Code runs in the following forms:

  • CLI terminal tool (primary form)
  • Desktop app (Mac/Windows)
  • Web app (claude.ai/code)
  • IDE extensions (VS Code, JetBrains)
  • Chrome extension integration (src/commands/chrome/)
  • Mobile (src/commands/mobile/)

README copy captured when this server was indexed; it may lag the repository. The original is authoritative.

Questions

About Claude Code Leak

How do I install Claude Code Leak?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Claude Code Leak safe to use with an AI agent?

Its trust score is 34 out of 100 (low). It passes 1 of the 1 static security checks run so far. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Claude Code Leak still maintained?

Conduid hasn't recorded a commit date for this repository yet. Check the repository directly for recent activity.