PolyMind Quick Start Guide

PolyMind is a feature-rich AI assistant that supports mainstream language models, providing environment awareness and powerful tool scheduling capabilities for the DevStation distribution.

Quick Start

1. Download and Installation

PolyMind was first released in openEuler-24.03-LTS-SP3 version and is pre-integrated in the DevStation platform.

Install the package:

sudo dnf install -y polymind

After successful installation, click the icon from the menu bar and launch the application.

2. Configure Model

After first launch, you need to configure the model provider:

Select model service provider
Automatically add the API address corresponding to the model service provider
Add your API key (such as OpenAI, DeepSeek, etc.) or configure local Ollama.
Next

**After configuring the model provider, further configure the model:**

Select model
Verify model (optional)
Next

**After configuration is complete, click Start Using.**

3. Start Conversation

Agent Configuration: Support creating, customizing and importing agents
Start Conversation:
- Select agent
- Enter message
- Press Enter to send
Settings: Support custom agent parameters, MCP services, search enhancement, etc.
Configuration Bar: Provide model parameters, MCP services, search enhancement and other configurations

Core Features

Agent Support

Local Agents: Support creating and customizing agents.
A2A Support: Support Agent-to-Agent communication for multi-agent collaboration.

Multi-Model Support

Cloud Models: DeepSeek, OpenAI, Gemini, Anthropic, Silicon Flow, Tencent Hunyuan, etc.
Local Models: Built-in Ollama integration, manage local models without command line.

Tool Calling

Built-in Model Context Protocol (MCP) support, providing tools for code execution, web retrieval, file operations, etc.
Models can automatically call built-in tools during conversations for environment awareness and scheduling.

Search Enhancement

Support for Brave Search, Google, Bing, Baidu and other search engines.
Models can intelligently decide when to search and highlight search results.

Advanced Rendering

Complete Markdown rendering with code block syntax highlighting.
Support for Mermaid diagrams, images, Artifacts and other multimedia content.

Developer Guide

Environment Preparation

Ensure the following are installed:

Node.js >= 20.19.0
pnpm >= 10.11.0

Install Dependencies

pnpm install
pnpm run installRuntime

Windows Users: If you encounter symbolic link errors, enable "Developer Mode" or run with administrator privileges.

Start Development Server

pnpm run dev

Build Application

Execute the corresponding command for your target platform:

# Windows
pnpm run build:win

# macOS
pnpm run build:mac

# Linux
pnpm run build:linux

For more architecture options, refer to the scripts in package.json.

Project Structure

src/
├── main/          # Electron main process
├── renderer/      # Vue frontend
├── preload/       # Preload scripts
└── shared/        # Shared types and tools
docs/              # Design documentation
scripts/           # Build scripts
resources/         # Static resources

Frequently Asked Questions

1. White Screen on Startup

Check the developer tools (Ctrl+Shift+I) for errors. It may be due to incomplete dependency installation. Try re-running pnpm install.

2. Ollama Connection Failure

Ensure the Ollama service is running (default address http://localhost:11434). You can modify the Ollama endpoint in settings.

3. API Key Configuration

Add the corresponding provider in "Model Providers", enter the API key and save. Keys are stored in local encrypted configuration.

4. Tool Calling Not Working

Check if MCP services are enabled. View the "MCP Services" tab in settings to ensure required services are enabled.