MCPAdapt: MCP Server for Seamless Integration in Agentic Frameworks

MCPAdapt is an open-source project designed to facilitate easy integration of Model Context Protocol (MCP) servers into various agentic frameworks. It allows anyone to access and utilize over 650 MCP servers directly within their preferred agentic setup, enabling widespread deployment of AI tools and resources.

Overview: What is MCPAdapt?

MCPAdapt makes it effortless for developers and integrators to leverage the Model Context Protocol (MCP) in their projects. MCP is an open-source protocol introduced by Anthropic that simplifies the process of making tools and data available as "MCP servers." These servers can be used by supported MCP clients, such as Claude Desktop, Continue, and Cursor, enabling powerful AI functionality.

🔧 Core Features & MCP Capabilities

MCPAdapt offers several key features:

• Seamless Integration: Direct integration with agentic frameworks like Smolagents, Langchain, CrewAI, google-genai, and more.
• Tool Adaptation: Converts MCP servers into tools that can be seamlessly used within any chosen framework.
• Async Support: Supports asynchronous operations if the underlying framework does not.

⚙️ MCP Architecture & Protocol Implementation

MCPAdapt implements the Model Context Protocol to enable communication between MCP clients and servers. This protocol defines how data and commands are transmitted, ensuring that both client and server adhere to a common standard for interaction. The implementation details include:

• Command Execution: Commands sent from the agentic framework are executed on the MCP server.
• Data Flow: Data is transferred securely between the client and server through standardized requests and responses.

🚀 Getting Started with Installation

To get started, you can install MCPAdapt using pip or as part of Smolagents. Here’s how:

Smolagents

Since version 1.4.1, Smolagents includes MCPAdapt pre-integrated. You can add it like so:

uv add smolagents[mcp]

Other Frameworks

For other frameworks, you need to install mcpadapt along with the specific framework adapter:

# Using pip
pip install mcpadapt[langchain]

# Or using uv in Smolagents context
uv add mcpadapt[langchain]

You can also specify multiple frameworks by separating them with commas.

💡 Key Use Cases in AI Workflows

Use Case 1: Real-Time Data Analysis

Imagine a scenario where an AI needs to process and analyze data streamed from an MCP server. Here’s how it would work:

from mcp import StdioServerParameters
from mcpadapt.core import MCPAdapt
from mcpadapt.smolagents_adapter import SmolAgentsAdapter

with MCPAdapt(
    StdioServerParameters(command="uv", args=["run", "src/analyze_data.py"]),
    SmolAgentsAdapter(),
) as tools:
    # Process and analyze data from the MCP server in real-time
    ...

Use Case 2: Custom Prompt Generation

Developers can use MCPAdapt to integrate custom prompt generation into their AI applications, allowing for dynamic and context-aware interactions.

from mcp import StdioServerParameters
from mcpadapt.core import MCPAdapt
from mcpadapt.langchain_adapter import LangChainAdapter

with MCPAdapt(
    StdioServerParameters(command="uv", args=["run", "src/generate_prompts.py"]),
    LangChainAdapter(),
) as tools:
    # Generate custom prompts based on user inputs and MCP server data
    ...

🔌 Integration with MCP Clients

MCPAdapt is compatible with various MCP clients, ensuring that developers can choose the right tool for their needs. Below is a compatibility matrix:

MCP Client	Resources	Tools	Prompts
Claude Desktop	✅	✅	✕
Continue	✕	✕	❌
Cursor	✖	✓	✕

📊 Performance & Compatibility Matrix

To ensure smooth integration, MCPAdapt is designed to work seamlessly across different agentic frameworks. We provide performance metrics and compatibility notes in our documentation.

Example Configuration Sample

Here’s an example of how you might configure your mcpServers section:

{
  "mcpServers": {
    "exampleServer": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-example"],
      "env": {
        "API_KEY": "your-api-key"
      }
    }
  }
}

🛠️ Advanced Configuration & Security

For advanced users, MCPAdapt offers several configuration options and security best practices. Ensure that your MCP servers are properly secured to prevent unauthorized access.

Security Considerations

When using MCP servers over SSE (Server-Sent Events), always ensure:

• TLS Encryption: Encrypted communication is a must.
• Source Verification: Only connect to trusted sources.
• Rate Limiting: Implement rate limiting to prevent abuse of your resources.

❓ Frequently Asked Questions (FAQ)

Q1: How do I integrate MCPAdapt with my smolagents project?

Integrating MCPAdapt into Smolagents is straightforward. Use the following command:

uv add smolagents[mcp]

Q2: Can I use MCPAdapt with Google GenAI servers?

Yes, you can adapt Google GenAI servers using MCPAdapt’s google_genai_adapter. Add it to your project as follows:

pip install mcpadapt[google_genai]

Q3: Is the synchronization between framework and MCP server real-time?

MCPAdapt supports real-time communication through asynchronous operations. Ensure that both frameworks support this feature.

Q4: How do I test an adapter configuration?

You can run tests by adapting tools from different servers and verifying their functionality within your agentic workflow.

Q5: Are there any specific environment variables required for MCPAdapt?

Yes, certain environment variables might be needed depending on the server you are integrating. Always refer to the official documentation of the server being used.

👨‍💻 Development & Contribution Guidelines

Contributing to MCPAdapt is easy and welcoming. Follow these steps:

1. Create a New Module: Add an adapter for your preferred agentic framework.
2. Implement Adapters: Write functions to convert MCP tools into usable framework tools.
3. Run Tests: Ensure that your code works as expected before submitting a Pull Request.

🌐 MCP Ecosystem & Resources

Explore more about the Model Context Protocol and its ecosystem at our official documentation:

• MCPAdapt Documentation
• Model Context Protocol Introduction

Join us in building a future where AI and tools work seamlessly together.

---

MCP Protocol Flow Diagram

graph TD
    A[AI Application] -->|MCP Client| B[MCP Protocol]
    B --> C[MCP Server]
    C --> D[Data Source/Tool]
    style A fill:#e1f5fe
    style C fill:#f3e5f5
    style D fill:#e8f5e8

MCP Client Compatibility Matrix

MCP Client	Resources	Tools	Prompts
Claude Desktop	✅	✅	✅ (Full)
Continue	❌	✅	❌ (Partial)
Cursor	✖	✓	✕

MCP Configuration Sample

{
  "mcpServers": {
    "mathServer": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-math"],
      "env": {
        "API_KEY": "your-api-key"
      }
    },
    "dataAnalysisServer": {
      "command": "npm",
      "args": ["run", "serve"]
    }
  }
}