Deepseek R1 Reasoner

🚀 DeepSeek r1 Inference Extension for Locally Run Models

This project offers a technical proxy system that relies entirely on local operations. It achieves this by integrating multiple models without the need for remote model calls.

🚀 Quick Start

✨ Features

• The system uses the DeepSeek r1 model for inference and can be combined with a tool - calling model. For instance, in the "Goose Project" agent, the Goose Agent can be used.
• DeepSeek r1 and other inference models usually struggle with tool - calling. However, this issue can be resolved by combining them with a model specifically designed for tool - calling.
• It's a general MCP server. As long as Ollama is running and all models are local, it can work with any agent system.

📦 Installation

npx -y deepseek-reasoner-mcp

💻 Usage Examples

This extension is applicable to any agent system.

# The system uses the DeepSeek r1 model for inference and can be combined with a tool - calling model.
# For example, in the "Goose Project" agent, use the Goose Agent as shown in the image below.
![image](https://github.com/user-attachments/assets/f017ece0-f352-4c24-8fa2-7faba4578fad)

# To start the extension, run the following command.
npx -y deepseek-reasoner-mcp

# This extension can be used with any agent system.

```markdown