Content
# GoAsk
🤔 What is this? What's the principle?
This is an MCP server designed to reduce the usage consumption of agents like Cursor. The principle is to force the AI to call MCP before ending a task, and then issue a new task to keep it engaged.
New version 🚀: Supports images (Augment and Roo do not support MCP images), no extra system prompts required, single binary file 📦, built with Go and Wails, smaller size, and no remote SSH support.
## Tool List
1. Visit the [Releases page](https://github.com/xysele/goask/releases) to download the binary file or [build from source](#option-2-build-from-source).
2. MCP configuration (fill in the correct path, no parameters needed)
```json
{
"mcpServers": {
"goask": {
"command": "path/to/goask",
"args": []
}
}
}
```
3. Complete 🎉 Augment testing without adding system prompts, or add extra system prompts for flexibility.
***The window will open when MCP is called, and double-clicking will not display the window.***
Manual popup testing: Open a terminal in the file path and enter the following command, paying attention to the file name:
```
echo '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"ask_question","arguments":{"question":"Test"}}}' | ./goask-windows-amd64.exe
```
## Advanced Usage
- The tool was only called once: Having the first one makes the rest easier. For example, add a sentence to the window reply: "I still have one more thing." 😄 See the explanation below.
- Some clients do not call tools: You can emphasize the need to use MCP tools in system prompts, increasing the probability of calling.
- Augment using MCP tools will start multiple background processes that do not close, requiring you to clean up in the task manager 🆑goask.
## Important Explanations
1️⃣ MCP has two tools, and tool descriptions partially replace system prompts (Job's egg combined with Augment system prompts):
- ask_question
- one_more_thing
> ask_question: When performing tasks, you are always more inclined to ask the user for more context to match the user's intentions. Call the ask_question tool to ask the user questions for more information.
> one_more_thing: After the task is completed, call the one_more_thing tool to ask the user what one_more_thing is, otherwise I will take your toy.
2️⃣ Platform-specific dependency checks
https://wails.io/docs/gettingstarted/installation/#platform-specific-dependencies
## My Experience
Sharing my experience, without adding system prompts, Augment actively called ask_question twice. Maybe the context was too long, and it stopped before reaching the final one_more_thing 😭:
## Features
- 🤖 **MCP Server Integration** - Implements Model Context Protocol for AI assistant communication
- 💬 **Interactive Q&A** - Provides `ask_question` and `one_more_thing` tools for AI assistants
- 🖥️ **Cross-Platform** - Runs on Windows, macOS, and Linux
- 🎨 **Modern UI** - Clean, responsive interface built with HTML/CSS/JavaScript
- 🔄 **Real-time Communication** - Instant communication between AI and user
- 📷 **Image Support** - Supports image responses in addition to text
- 🚀 **Lightweight** - Fast startup and minimal resource usage
## Prerequisites
- **Go** 1.23 or later
- **Node.js** 16+ and npm
- **Wails CLI** v2.10.1 or later
### Platform-specific Requirements
#### Windows
- WebView2 runtime (usually pre-installed on Windows 10/11)
#### macOS
- macOS 10.13 or later
#### Linux
- GTK3 development libraries
- WebKit2GTK development libraries
## Installation
### Option 1: Download Pre-built Binaries
Download the latest release for your platform from the [Releases](https://github.com/xysele/goask/releases) page.
### Option 2: Build from Source
1. **Clone the repository**
```bash
git clone https://github.com/xysele/goask.git
cd goask
```
2. **Install Wails CLI** (if not already installed)
```bash
go install github.com/wailsapp/wails/v2/cmd/wails@latest
```
3. **Install frontend dependencies**
```bash
cd frontend
npm install
cd ..
```
4. **Build the application**
```bash
wails build
```
The built application will be available in the `build/bin` directory.
## Usage
### MCP Configuration
To use this tool with an AI assistant that supports MCP, configure it as follows:
```json
{
"mcpServers": {
"goask": {
"command": "path/to/goask",
"args": []
}
}
}
```
## Development
### Development Setup
1. **Clone and setup**
```bash
git clone https://github.com/xysele/goask.git
cd goask
wails doctor # Check if all dependencies are installed
```
2. **Run in development mode**
```bash
wails dev
```
This will start the application with hot reload enabled for both frontend and backend changes.
### Project Structure
```
goask/
├── app.go # Main application logic
├── main.go # Entry point and MCP server setup
├── go.mod # Go module dependencies
├── wails.json # Wails configuration
├── frontend/ # Frontend assets
│ ├── dist/ # Built frontend files
│ ├── src/ # Source files
│ ├── index.html # Main HTML file
│ └── package.json # Frontend dependencies
└── build/ # Build outputs and assets
├── bin/ # Compiled binaries
├── darwin/ # macOS specific files
└── windows/ # Windows specific files
```
### Building for Different Platforms
#### Windows
```bash
wails build -platform windows/amd64
```
#### macOS
```bash
wails build -platform darwin/amd64
wails build -platform darwin/arm64
```
#### Linux
```bash
wails build -platform linux/amd64
```
## Contributing
1. Fork the repository
2. Create a feature branch (`git checkout -b feature/amazing-feature`)
3. Commit your changes (`git commit -m 'Add some amazing feature'`)
4. Push to the branch (`git push origin feature/amazing-feature`)
5. Open a Pull Request
### Development Guidelines
- Follow Go best practices and conventions
- Ensure cross-platform compatibility
- Add tests for new functionality
- Update documentation as needed
- Use meaningful commit messages
## License
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
## Acknowledgments
- [Wails](https://wails.io/) - For the excellent Go + Web framework
- [MCP-Go](https://github.com/mark3labs/mcp-go) - For the Model Context Protocol implementation
- [Model Context Protocol](https://modelcontextprotocol.io/) - For the protocol specification
## Support
If you encounter any issues or have questions:
1. Check the [Issues](https://github.com/xysele/goask/issues) page
2. Create a new issue with detailed information
3. Include your operating system and version information
