https://github.com/gcwing/bitfun
BitFun is a next-generation AI assistant with built-in Code Agent and Cowork Agent. It has memory, personality, and the ability to evolve over time. You can remotely control the desktop through mobile QR pairing or Telegram / Feishu bots, send instructions, and watch each execution step in real time while the Agent keeps working in the background.
https://github.com/gcwing/bitfun
agent agent-teams agentic-os ai ai-coding ai-ide ai-os cowork human-computer-interaction ide openclaw vibe-coding
Last synced: 2 days ago
JSON representation
BitFun is a next-generation AI assistant with built-in Code Agent and Cowork Agent. It has memory, personality, and the ability to evolve over time. You can remotely control the desktop through mobile QR pairing or Telegram / Feishu bots, send instructions, and watch each execution step in real time while the Agent keeps working in the background.
- Host: GitHub
- URL: https://github.com/gcwing/bitfun
- Owner: GCWing
- License: mit
- Created: 2026-02-02T09:52:10.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2026-04-17T09:28:02.000Z (9 days ago)
- Last Synced: 2026-04-17T11:35:31.288Z (9 days ago)
- Topics: agent, agent-teams, agentic-os, ai, ai-coding, ai-ide, ai-os, cowork, human-computer-interaction, ide, openclaw, vibe-coding
- Language: TypeScript
- Homepage:
- Size: 49.8 MB
- Stars: 582
- Watchers: 2
- Forks: 67
- Open Issues: 13
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
- Agents: AGENTS.md
Awesome Lists containing this project
README
[中文](README.zh-CN.md) | **English**
**AI assistant with personality and memory**
Hand over the work, keep the life
AI Agent workspace for coding and knowledge work
[](https://github.com/GCWing/BitFun/releases)
[](https://openbitfun.com/)
[](https://github.com/GCWing/BitFun/blob/main/LICENSE)
[](https://github.com/GCWing/BitFun)
---
## Introduction
In the age of AI, true human-machine collaboration is not just a ChatBox, but a partner that understands you, works with you, keeps evolving, and gets things done anytime, anywhere. That is where BitFun begins.
BitFun is a next-generation AI assistant with built-in **Code Agent** and **Cowork Agent**. It has memory, personality, and the ability to evolve over time. You can remotely control the desktop through mobile QR pairing or Telegram / Feishu bots, send instructions, and watch each execution step in real time while the Agent keeps working in the background.
Built with **Rust + TypeScript** for an ultra-lightweight, fluid, cross-platform experience.
---
## Dual Modes
BitFun provides two modes for different kinds of work:
- **Assistant Mode**: warmer, preference-aware, and backed by long-term memory. Best for ongoing collaboration, such as maintaining a project or preserving your writing and design style.
- **Professional Mode**: token-efficient, execution-first, and context-clean. Best for short, focused tasks like fixing a bug or updating a small feature.
---
## Remote Control
With QR pairing, your phone instantly becomes a remote command center for the desktop Agent. Send one message and the desktop AI starts working right away.
The desktop generates a QR code, and the mobile browser opens the remote interface after scanning it, with no app installation required.
Besides mobile QR pairing, BitFun also supports Telegram / Feishu bots for remote instructions and real-time progress tracking.
| Feature | Description |
|---|---|
| **QR Pairing** | Scan a QR code generated by the desktop, complete key exchange, and bind a long-lived connection |
| **Full Control** | View sessions, switch modes, send instructions, and control the desktop workflow remotely |
| **Real-time Streaming** | Every Agent step and tool call can be viewed live on your phone |
## Agent System
| Agent | Role | Core Capabilities |
|---|---|---|
| **Personal Assistant** (Beta) | Your dedicated AI companion | Long-term memory and personality settings; can orchestrate Code / Cowork / custom Agents on demand, and continuously evolve |
| **Code Agent** | Coding assistant | Four modes: Agentic (autonomous read / edit / run / verify) / Plan (plan first, then execute) / Debug (instrumentation to root cause) / Review (repository-aware code review) |
| **Cowork Agent** | Knowledge work assistant | Built-in PDF / DOCX / XLSX / PPTX handling, and can fetch and extend suitable capability packages from the Skill marketplace |
| **Custom Agent** | Domain specialist | Quickly define a domain-specific Agent with Markdown |
---
## Ecosystem
> It keeps growing.
Mini Apps emerge from conversations, Skills evolve in the community, and Agents improve through collaboration.
| Layer | Description |
|---|---|
| **Mini Apps** | Generate runnable interfaces from a prompt and package them into desktop apps with one click |
| **Skill Marketplace** | Install community capability packs so Agents can learn new skills quickly |
| **MCP Protocol** | Connect external tools and resources to extend Agent capabilities beyond the local system |
| **Custom Agents** | Define roles, memory, and capability boundaries with Markdown |
| **ACP Protocol (WIP)** | A structured multi-Agent communication standard for interoperating with mainstream AI tools |
---
## Platform Support
The project is built with Rust + TypeScript for cross-platform reuse, keeping your Agent available wherever you work.
| Form Factor | Supported Platforms | Status |
|---|---|---|
| **Desktop** | Windows, macOS, Linux | ✅ Supported (Tauri) |
| **Remote Control** | Mobile browser, Telegram, Feishu | ✅ Supported |
---
## Quick Start
### Download and Use
Download the latest desktop installer from [Releases](https://github.com/GCWing/BitFun/releases). After installation, configure your model and start using BitFun.
> CLI, Server, and native mobile apps are still in planning or development. Desktop and remote control are already supported.
### Build from Source
Make sure you have the following prerequisites installed:
- Node.js (LTS recommended)
- pnpm
- Rust toolchain (install via [rustup](https://rustup.rs/))
- [Tauri prerequisites](https://v2.tauri.app/start/prerequisites/) for desktop development
**Windows only**: The desktop build links against a **prebuilt** OpenSSL (no OpenSSL source compile). You do **not** need to download the ZIP by hand: the first time OpenSSL is required, tooling fetches [FireDaemon OpenSSL 3.5.5](https://download.firedaemon.com/FireDaemon-OpenSSL/openssl-3.5.5.zip) into **`.bitfun/cache/`** and later runs reuse that cache. **`pnpm run desktop:dev`** and all **`pnpm run desktop:build*`** scripts run `ensure-openssl-windows.mjs` (builds use `desktop-tauri-build.mjs`). **If you compile with plain `cargo`** (without those pnpm entrypoints), run **`node scripts/ensure-openssl-windows.mjs`** once from the repo root first — it performs the same download and prints **`OPENSSL_*`** lines for PowerShell. Override with `OPENSSL_DIR` pointing at the **`x64`** folder from the ZIP, or `BITFUN_SKIP_OPENSSL_BOOTSTRAP=1` and your own `OPENSSL_*`.
```bash
# Install dependencies
pnpm install
# Run desktop app in development mode
pnpm run desktop:dev
# Build desktop app
pnpm run desktop:build
```
For more details, see the [Chinese Contributing Guide](./CONTRIBUTING_CN.md).
### Linux Build
#### Prerequisites
Install system dependencies:
```bash
# Debian/Ubuntu
sudo apt install libwebkit2gtk-4.1-dev build-essential libgtk-3-dev \
libayatana-appindicator3-dev librsvg2-dev patchelf
```
See [docs/linux-setup.md](docs/linux-setup.md) for other distributions (Arch, Fedora, etc.).
#### Build
```bash
pnpm install
pnpm run desktop:build:linux
```
Output will be in `src/apps/desktop/target/release/bundle/` (`.deb`, `.rpm`, `.AppImage`).
---
## Contributing
We welcome great ideas and code contributions. We are maximally accepting of AI-generated code. Please submit PRs to the `dev` branch first; we will periodically review and sync to the main branch.
Key contribution areas we focus on:
1. Contributing good ideas and creativity in features, interaction, and visual design via Issues
2. Optimizing the Agent system and overall quality
3. Improving system stability and foundational capabilities
4. Expanding the ecosystem, including Skills, MCP, LSP plugins, and support for vertical development scenarios
---
## Disclaimer
1. This project is built in spare time for exploring and researching next-generation human-machine collaboration, not for commercial profit.
2. 97%+ of this project was built with Vibe Coding. Feedback on code issues is also welcome, and refactoring or optimization through AI is encouraged.
3. This project depends on and references many open-source projects. Thanks to all open-source authors. **If your rights are affected, please contact us for rectification.**
---
The world is being rewritten, and this time, we are all holding the pen.