# Claude Code Haha

> A local-first cross-platform desktop workspace for Claude Code and AI agents with multi-session support, multi-model switching, Git worktrees, skill marketplace, and IM integrations.

Claude Code Haha is an open-source desktop application built by NanmiCoder that wraps Claude Code and other AI coding agents in a full-featured GUI for macOS, Windows, and Linux. Released under the MIT License, it is freely available on GitHub and actively maintained with regular releases. The project is written in TypeScript using Electron for the desktop shell and React + Vite for the UI.

## What It Is

Claude Code Haha is a desktop workspace that sits on top of Claude Code (Anthropic's CLI coding agent) and extends it with a rich graphical interface. Rather than running Claude Code in a bare terminal, users get a tabbed multi-session environment, a visual model picker, a built-in browser preview, GUI permission approval dialogs, and a skill marketplace — all in a single installable app. It supports any model provider, not just Anthropic: Claude, ChatGPT, Grok, DeepSeek, Kimi, Zhipu GLM, and local endpoints via LM Studio or Ollama.

## Core Feature Set

The README lists an extensive set of built-in capabilities:

- **Multi-session workspace**: tabs, project switching, terminal entry, and session history with a resizable sidebar
- **Global search**: Cmd+K search across every session
- **Branch / Worktree launch**: choose a repository branch and optionally use an isolated Git Worktree
- **Diff review**: syntax-highlighted file-by-file diff with per-turn undo
- **Built-in browser preview**: renders the edited page inside the app with cookies and login state preserved
- **Five permission modes**: from "ask every time" to "skip permissions" — all approved in the GUI
- **Visual MCP manager**: add and configure MCP servers (STDIO / Streamable HTTP / SSE) with project, shared, or global scope
- **Visual SubAgent manager**: create and tune SubAgents with model, tools, and permission mode
- **Model trace**: every model request logged locally with status and timing for debugging
- **Computer Use**: agent-controlled screenshots, clicks, and typing after authorization
- **Skill marketplace**: discover and install third-party skills from ClawHub / SkillHub
- **Image generation**: generate and edit images via ChatGPT, Grok, or any OpenAI-compatible Images API
- **Desktop pets**: animated companions (Dada, Huhu, Bubu, Huihui) that react to task state
- **H5 remote access**: scan a QR code to continue a session in a phone browser
- **IM integration**: chat and approve actions through Telegram, Feishu, WeChat, DingTalk, and WhatsApp
- **Scheduled tasks and usage stats**: run planned tasks and track local token usage trends
- **Six colour themes**: white, paper, warm classic, celadon, ink night, and ink blue

## Architecture and Tech Stack

The project uses a layered architecture documented in the repository internals section:

| Layer | Technology |
|---|---|
| Language | TypeScript |
| Desktop shell | Electron |
| Desktop UI | React + Vite |
| Local runtime | Bun |
| Terminal UI | React + Ink |
| CLI parsing | Commander.js |
| API | Anthropic SDK |
| Protocols | MCP, LSP |

The CLI can also be run directly from source with `bun install` and `./bin/claude-haha`, making it accessible to developers who want to debug the underlying server or contribute.

## Update: v0.5.3

The latest release is **v0.5.3**, published on 2026-08-04. The repository was last pushed on 2026-08-06, indicating active development. The project has accumulated over 13,900 GitHub stars and 8,500 forks since its creation in March 2026, signaling rapid community adoption. The GitHub topics include `agent`, `claude-code`, `computer-use`, `mcp`, and `electron`, reflecting the breadth of the feature surface.

## Setup Path

Installation follows three steps: download the macOS, Windows, or Linux installer from the GitHub Releases page; launch the app and configure a model provider, API key, and default model in Settings; then start a session. Public macOS releases are signed and notarized; unsigned Windows builds may trigger SmartScreen and require a manual "Run anyway" step. Full documentation is available at cchaha.ai.

## Features
- Multi-session workspace with tabs and project switching
- Global search across all sessions (Cmd+K)
- Branch and Git Worktree launch
- Syntax-highlighted file-by-file diff review with per-turn undo
- Built-in browser preview with cookies and login state
- Five permission modes with GUI approval dialogs
- Multi-model support: Claude, ChatGPT, Grok, DeepSeek, Kimi, Zhipu GLM, LM Studio, Ollama
- Image generation via ChatGPT, Grok, or OpenAI-compatible Images API
- Visual MCP server manager (STDIO, Streamable HTTP, SSE)
- Visual SubAgent manager
- Model trace with local logging, search, and filter
- Computer Use: agent-controlled screenshots, clicks, and typing
- Skill marketplace (ClawHub / SkillHub)
- Session activity panel for task progress and SubAgents
- Desktop pets (Dada, Huhu, Bubu, Huihui)
- H5 remote access via QR code for phone browser
- IM integration: Telegram, Feishu, WeChat, DingTalk, WhatsApp
- Scheduled tasks and local token usage stats
- Six colour themes with system light/dark follow

## Integrations
Claude (Anthropic), ChatGPT (OpenAI), Grok, DeepSeek, Kimi, Zhipu GLM, LM Studio, Ollama, MCP (Model Context Protocol), LSP (Language Server Protocol), Telegram, Feishu (Lark), WeChat, DingTalk, WhatsApp, ClawHub / SkillHub, Git / Worktrees

## Platforms
WINDOWS, MACOS, LINUX, WEB, API, CLI

## Pricing
Open Source

## Version
v0.5.3

## Links
- Website: https://cchaha.ai/
- Documentation: https://cchaha.ai/en/start
- Repository: https://github.com/NanmiCoder/cc-haha
- EveryDev.ai: https://www.everydev.ai/tools/claude-code-haha