{
"markdown": "\n\n# Claude Code — Leaked Source\n\n**The full source code of Anthropic's Claude Code CLI, leaked on March 31, 2026**\n\n[](#tech-stack)\n[](#tech-stack)\n[](#tech-stack)\n[](#directory-structure)\n[](#-explore-with-mcp-server)\n[](https://www.npmjs.com/package/warrioraashuu-codemaster)\n[](https://twitter.com/intent/follow?screen_name=warrioraashuu)\n\n> The original unmodified leaked source is preserved in the [`backup` branch](https://github.com/codeaashu/claude-code/tree/backup).\n\n
\n\n---\n\n \n
\n\n \n`The X-Ray Intelligence Engine for Your Codebase.`\n \n\n \n\nUnderstand Any Codebaserepoxray.2clabs.tech \n \n \n
\n\n`Spend less time reading code, more time building.`\n\n
\nStep-by-step setup \n\n\nVS Code / Cursor / Claude Desktop config \n\n### Available tools & prompts\n\n| Tool | Description |\n|------|-------------|\n| `list_tools` | List all ~40 agent tools with source files |\n| `list_commands` | List all ~50 slash commands with source files |\n| `get_tool_source` | Read full source of any tool (e.g. BashTool, FileEditTool) |\n| `get_command_source` | Read source of any slash command (e.g. review, mcp) |\n| `read_source_file` | Read any file from `src/` by path |\n| `search_source` | Grep across the entire source tree |\n| `list_directory` | Browse `src/` directories |\n| `get_architecture` | High-level architecture overview |\n\n| Prompt | Description |\n|--------|-------------|\n| `explain_tool` | Deep-dive into how a specific tool works |\n| `explain_command` | Understand a slash command's implementation |\n| `architecture_overview` | Guided tour of the full architecture |\n| `how_does_it_work` | Explain any subsystem (permissions, MCP, bridge, etc.) |\n| `compare_tools` | Side-by-side comparison of two tools |\n\n**Try asking:** *\"How does the BashTool work?\"* · *\"Search for where permissions are checked\"* · *\"Show me the /review command source\"*\n\n### Custom source path / Remove\n\n```bash\n# Custom source location\nclaude mcp add claude-code-explorer -e CLAUDE_CODE_SRC_ROOT=/path/to/src -- node /path/to/mcp-server/dist/index.js\n\n# Remove\nclaude mcp remove claude-code-explorer\n```\n\n---\n\n## Directory Structure\n\n```\nsrc/\n├── main.tsx # Entrypoint — Commander.js CLI parser + React/Ink renderer\n├── QueryEngine.ts # Core LLM API caller (~46K lines)\n├── Tool.ts # Tool type definitions (~29K lines)\n├── commands.ts # Command registry (~25K lines)\n├── tools.ts # Tool registry\n├── context.ts # System/user context collection\n├── cost-tracker.ts # Token cost tracking\n│\n├── tools/ # Agent tool implementations (~40)\n├── commands/ # Slash command implementations (~50)\n├── components/ # Ink UI components (~140)\n├── services/ # External service integrations\n├── hooks/ # React hooks (incl. permission checks)\n├── types/ # TypeScript type definitions\n├── utils/ # Utility functions\n├── screens/ # Full-screen UIs (Doctor, REPL, Resume)\n│\n├── bridge/ # IDE integration (VS Code, JetBrains)\n├── coordinator/ # Multi-agent orchestration\n├── plugins/ # Plugin system\n├── skills/ # Skill system\n├── server/ # Server mode\n├── remote/ # Remote sessions\n├── memdir/ # Persistent memory directory\n├── tasks/ # Task management\n├── state/ # State management\n│\n├── voice/ # Voice input\n├── vim/ # Vim mode\n├── keybindings/ # Keybinding configuration\n├── schemas/ # Config schemas (Zod)\n├── migrations/ # Config migrations\n├── entrypoints/ # Initialization logic\n├── query/ # Query pipeline\n├── ink/ # Ink renderer wrapper\n├── buddy/ # Companion sprite (Easter egg 🐣)\n├── native-ts/ # Native TypeScript utils\n├── outputStyles/ # Output styling\n└── upstreamproxy/ # Proxy configuration\n```\n\n---\n\n## Architecture\n\n### 1. Tool System\n\n> `src/tools/` — Every tool Claude can invoke is a self-contained module with its own input schema, permission model, and execution logic.\n\n| Tool | Description |\n|---|---|\n| **File I/O** | |\n| `FileReadTool` | Read files (images, PDFs, notebooks) |\n| `FileWriteTool` | Create / overwrite files |\n| `FileEditTool` | Partial modification (string replacement) |\n| `NotebookEditTool` | Jupyter notebook editing |\n| **Search** | |\n| `GlobTool` | File pattern matching |\n| `GrepTool` | ripgrep-based content search |\n| `WebSearchTool` | Web search |\n| `WebFetchTool` | Fetch URL content |\n| **Execution** | |\n| `BashTool` | Shell command execution |\n| `SkillTool` | Skill execution |\n| `MCPTool` | MCP server tool invocation |\n| `LSPTool` | Language Server Protocol integration |\n| **Agents & Teams** | |\n| `AgentTool` | Sub-agent spawning |\n| `SendMessageTool` | Inter-agent messaging |\n| `TeamCreateTool` / `TeamDeleteTool` | Team management |\n| `TaskCreateTool` / `TaskUpdateTool` | Task management |\n| **Mode & State** | |\n| `EnterPlanModeTool` / `ExitPlanModeTool` | Plan mode toggle |\n| `EnterWorktreeTool` / `ExitWorktreeTool` | Git worktree isolation |\n| `ToolSearchTool` | Deferred tool discovery |\n| `SleepTool` | Proactive mode wait |\n| `CronCreateTool` | Scheduled triggers |\n| `RemoteTriggerTool` | Remote trigger |\n| `SyntheticOutputTool` | Structured output generation |\n\n### 2. Command System\n\n> `src/commands/` — User-facing slash commands invoked with `/` in the REPL.\n\n| Command | Description | | Command | Description |\n|---|---|---|---|---|\n| `/commit` | Git commit | | `/memory` | Persistent memory |\n| `/review` | Code review | | `/skills` | Skill management |\n| `/compact` | Context compression | | `/tasks` | Task management |\n| `/mcp` | MCP server management | | `/vim` | Vim mode toggle |\n| `/config` | Settings | | `/diff` | View changes |\n| `/doctor` | Environment diagnostics | | `/cost` | Check usage cost |\n| `/login` / `/logout` | Auth | | `/theme` | Change theme |\n| `/context` | Context visualization | | `/share` | Share session |\n| `/pr_comments` | PR comments | | `/resume` | Restore session |\n| `/desktop` | Desktop handoff | | `/mobile` | Mobile handoff |\n\n### 3. Service Layer\n\n> `src/services/` — External integrations and core infrastructure.\n\n| Service | Description |\n|---|---|\n| `api/` | Anthropic API client, file API, bootstrap |\n| `mcp/` | Model Context Protocol connection & management |\n| `oauth/` | OAuth 2.0 authentication |\n| `lsp/` | Language Server Protocol manager |\n| `analytics/` | GrowthBook feature flags & analytics |\n| `plugins/` | Plugin loader |\n| `compact/` | Conversation context compression |\n| `extractMemories/` | Automatic memory extraction |\n| `teamMemorySync/` | Team memory synchronization |\n| `tokenEstimation.ts` | Token count estimation |\n| `policyLimits/` | Organization policy limits |\n| `remoteManagedSettings/` | Remote managed settings |\n\n### 4. Bridge System\n\n> `src/bridge/` — Bidirectional communication layer connecting IDE extensions (VS Code, JetBrains) with the CLI.\n\nKey files: `bridgeMain.ts` (main loop) · `bridgeMessaging.ts` (protocol) · `bridgePermissionCallbacks.ts` (permission callbacks) · `replBridge.ts` (REPL session) · `jwtUtils.ts` (JWT auth) · `sessionRunner.ts` (session execution)\n\n### 5. Permission System\n\n> `src/hooks/toolPermission/` — Checks permissions on every tool invocation.\n\nPrompts the user for approval/denial or auto-resolves based on the configured permission mode: `default`, `plan`, `bypassPermissions`, `auto`, etc.\n\n### 6. Feature Flags\n\nDead code elimination at build time via Bun's `bun:bundle`:\n\n```typescript\nimport { feature } from 'bun:bundle'\n\nconst voiceCommand = feature('VOICE_MODE')\n ? require('./commands/voice/index.js').default\n : null\n```\n\nNotable flags: `PROACTIVE` · `KAIROS` · `BRIDGE_MODE` · `DAEMON` · `VOICE_MODE` · `AGENT_TRIGGERS` · `MONITOR_TOOL`\n\n---\n\n## Key Files\n\n| File | Lines | Purpose |\n|------|------:|---------|\n| `QueryEngine.ts` | ~46K | Core LLM API engine — streaming, tool loops, thinking mode, retries, token counting |\n| `Tool.ts` | ~29K | Base types/interfaces for all tools — input schemas, permissions, progress state |\n| `commands.ts` | ~25K | Command registration & execution with conditional per-environment imports |\n| `main.tsx` | — | CLI parser + React/Ink renderer; parallelizes MDM, keychain, and GrowthBook on startup |\n\n---\n\n## Tech Stack\n\n| Category | Technology |\n|---|---|\n| Runtime | [Bun](https://bun.sh) |\n| Language | TypeScript (strict) |\n| Terminal UI | [React](https://react.dev) + [Ink](https://github.com/vadimdemedes/ink) |\n| CLI Parsing | [Commander.js](https://github.com/tj/commander.js) (extra-typings) |\n| Schema Validation | [Zod v4](https://zod.dev) |\n| Code Search | [ripgrep](https://github.com/BurntSushi/ripgrep) (via GrepTool) |\n| Protocols | [MCP SDK](https://modelcontextprotocol.io) · LSP |\n| API | [Anthropic SDK](https://docs.anthropic.com) |\n| Telemetry | OpenTelemetry + gRPC |\n| Feature Flags | GrowthBook |\n| Auth | OAuth 2.0 · JWT · macOS Keychain |\n\n---\n\n## Design Patterns\n\n\nParallel Prefetch — Startup optimization \n\n\nLazy Loading — Deferred heavy modules \n\n\nAgent Swarms — Multi-agent orchestration \n\n\nSkill System — Reusable workflows \n\n\nPlugin Architecture — Extensibility \n\n---\n\n## GitPretty Setup\n\n\nShow per-file emoji commit messages in GitHub's file UI \n\n```bash\n# Apply emoji commits\nbash ./gitpretty-apply.sh .\n\n# Optional: install hooks for future commits\nbash ./gitpretty-apply.sh . --hooks\n\n# Push as usual\ngit push origin main\n```\n\n \n\n---\n\n## Contributing\n\nContributions to documentation, the MCP server, and exploration tooling are welcome. See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.\n\n> **Note:** The `src/` directory is the original leaked source and should not be modified.\n\n---\n\n## Disclaimer\n\nThis repository archives source code leaked from Anthropic's npm registry on **2026-03-31**. All original source code is the property of [Anthropic](https://www.anthropic.com). This is not an official release and is not licensed for redistribution. Contact [aashuu ✦](https://x.com/warrioraashuu) for any comments.\n\n---\n\n\n \n \n \n\n\n---\n\nCooked by aashuu ✦ (Ashutosh Kumar) ![\"RepoXray](\"https://repoxray.2clabs.tech/logo.png\"){kind=logo}
\n