Understanding Model Context Protocol (MCP)
The Model Context Protocol (MCP) is an open standard that aims to streamline how AI models, particularly Large Language Models (LLMs), connect with external data sources and tools. Think of it as a universal language that allows AI to access and utilize information from various systems in a standardized way.
Key Concepts of MCP
- Standardized Connections: MCP provides a consistent way for AI to interact with different servers, replacing the need for custom integrations for each data source.
- Contextual Awareness: MCP enables AI to access the specific data it needs to understand a situation or answer a query, rather than relying solely on its internal knowledge.
- Modular Architecture: MCP separates the AI application (host) from the data providers (servers), allowing for flexibility and extensibility.
MCP Server
In the context of MCP, a server is a component that exposes a specific data source or tool to AI applications. An MCP server:
- Provides access to data (e.g., a database, a file system).
- Offers tools or functionalities (e.g., search, data manipulation).
- Communicates with AI applications using the MCP standard.
How TigerGraph Plays a Role in the MCP Server Space
TigerGraph can power an MCP server, providing AI applications with access to rich, interconnected data and analytical capabilities.
Here’s how:
- Exposing Graph Data via MCP: TigerGraph can expose its graph data and query functionalities through an MCP server interface. This allows AI models to:
- Retrieve information about entities and their relationships: For example, an AI agent could use TigerGraph to find all customers connected to a specific transaction or identify relationships between different accounts in a financial network.
- Execute graph queries: AI can leverage TigerGraph’s GSQL query language to perform complex graph traversals and analytics, enabling it to answer questions that require understanding relationships within the data.
- Providing Context for AI Reasoning: By acting as an MCP server, TigerGraph can equip AI models with the contextual information they need to make more informed and accurate decisions. For instance, in a customer service application, an AI agent can use TigerGraph to access a customer’s interaction history, social connections, and purchase patterns to provide more personalized and helpful support.
- Enhancing AI Explainability: The graph-based structure of TigerGraph makes it easier to understand how AI arrived at a particular conclusion. By tracing the paths and relationships used by an AI agent, TigerGraph can improve the transparency and explainability of AI decision-making.
Use Cases
Here are some examples of how TigerGraph as an MCP server can be used:
- AI-Powered Customer Service: An AI assistant uses TigerGraph to access customer data and relationship information to provide personalized and context-aware support.
- Dynamic Fraud Detection: An AI agent leverages TigerGraph to analyze transaction networks and identify complex fraud patterns in real-time.
- Knowledge-Driven Applications: An AI system uses TigerGraph to query a knowledge graph and provide users with accurate and comprehensive answers.
By acting as an MCP server, TigerGraph can empower AI applications with the ability to understand and reason over complex relationships, leading to more intelligent and effective solutions.
Get Started
Prerequisites
To use TigerGraph-MCP, ensure you have the following prerequisites:
1. Python: version 3.10, 3.11, or 3.12.
2. TigerGraph: You need TigerGraph version 4.1 or later. You can set it up using one of these methods:
- Local Installation: Install and configure TigerGraph on your machine.
- TigerGraph Savanna: Use a managed instance of TigerGraph.
- Docker: Run TigerGraph in a containerized environment.
Installation Steps
Option 1: Install from PyPI
The simplest way to install TigerGraph-MCP is via PyPI. It is recommended to create a virtual environment first:
Shell
pip install tigergraph-mcp
Option 2: Build from Source
If you wish to explore or modify the code:
1. Install Poetry for dependency management.
2. Clone the repository:
Shell git clone https://github.com/TigerGraph-DevLabs/tigergraphx cd tigergraph-mcp
3. Setting up the Python environment with Poetry
Shell poetry env use python3.12 poetry install --with dev eval $(poetry env activate)
Using TigerGraph-MCP Tools
To utilize TigerGraph-MCP tools effectively, especially with GitHub Copilot Chat in VS Code, follow these steps:
1. Set Up GitHub Copilot Chat: Follow the official documentation to configure it.
2. Create a .env File: Include your OpenAI API key and TigerGraph connection details.
3. Configure VS Code: Create a .vscode/mcp.json file to set up the TigerGraph-MCP server.
4. Interact with the MCP Tool: Use GitHub Copilot to send commands and create schemas in TigerGraph.
Advanced Usage with CrewAI
For more complex interactions or custom workflows, consider using CrewAI or LangGraph. Examples are provided in the repository to help you get started with creating AI agents and managing workflows.
TigerGraph MCP server is open-source at: https://github.com/TigerGraph-DevLabs/tigergraph-mcp/tree/main.
Current Status
The TigerGraph MCP server is actively being developed, and we encourage you to contribute! Here are some current features and enhancements:
- Basic MCP Functionality: The server currently supports basic data retrieval and query execution through the MCP interface. You can view the list of currently supported features here.
- Ongoing Improvements: We are continuously working on enhancing the server’s capabilities. For details on our development roadmap, please visit here.
- Community Contributions: We welcome community feedback and contributions. If you have ideas for new features or improvements, please open an issue or submit a pull request on GitHub.
Follow the demo video below to give it a try here
