Fluent MCP

🚀 Introduction to the Fluent MCP Framework

The Fluent MCP (Model Context Protocol) Framework is a modern framework for building model context protocol servers, with a particular focus on artificial intelligence inference capabilities. It offers an efficient, secure, and easily extensible way to manage and run complex AI models.

🚀 Quick Start

What is the Fluent MCP Framework?

The Fluent MCP (Model Context Protocol) Framework is a modern framework for building model context protocol servers, specifically emphasizing artificial intelligence inference capabilities. It provides an efficient, secure, and easily extensible approach to managing and running complex AI models.

✨ Features

Core Architecture Patterns

Two - layer LLM Structure

• Embedded LLM: Responsible for handling internal complex reasoning and calculations.
• Consumable LLM: Used for external models (e.g., Claude), accessed via API.

Tool Separation

• Internal Tools: Restricted for use by the embedded LLM, enhancing system security.
• External Tools: Exposed to the consumable LLM for functionality expansion.

📦 Installation

Creating a New Server

Command - line Generation

# Use the command to generate the project structure
fluent-mcp new my_server
cd my_server
npm install

Programmatic Creation

from fluent_mcp import create_mcp_server

server = create_mcp_server(
    server_name="my_server",
    embedded_tools=[search_database, analyze_data],
    external_tools=[research_question],
    config={
        "provider": "ollama",
        "model": "llama2",
        "base_url": "http://localhost:11434",
        "api_key": "ollama"
    }
)

server.run()

💻 Usage Examples

Implementing Core Architecture Patterns

Registering Tools

from fluent_mcp.core.tool_registry import register_tool

@register_tool(name="search_database")
def search_database(query):
    # Query the database and return the result
    pass

@register_tool(name="analyze_data")
def analyze_data(data):
    # Analyze the data and return the result
    pass

Using Tools

from fluent_mcp.core.tool_registry import get_tool

# Get the registered tools
database_tool = get_tool("search_database")
data_analyzer = get_tool("analyze_data")

# Perform operations
resultDatabase = database_tool(query)
resultAnalysis = data_analyzer(data)

Tool Definitions in Prompts

Enhance control and security by specifying available tools in the prompt file:

---
name: math_tools
description: A prompt using math - related tools
model: gpt-4
temperature: 0.3
tools:
  - add_numbers
  - multiply_numbers
---
You are a math assistant capable of performing calculations.
Use the available tools to help solve math problems.

📚 Documentation

For detailed documentation, please refer to the docs directory:

• Getting Started
• Creating Tools
• Running a Server
• Embedded Reasoning
• Rate Limiting
• Self - Improving Systems
• Troubleshooting
• API Reference

Example Code

For more examples, please check the examples directory:

• Tool Registry Example
• Embedded Reasoning Example
• External Tools Example
• Reasoning External Tool Example
• Prompt Loader Example
• Prompt Tools Example

🔧 Technical Details

Development Information

Installing Dependencies

pip install -e ".[dev]"

Running Tests

pytest

Code Style

flake8
black .
isort .

📄 License

MIT