We are pleased to announce the release of TigerGraph MCP version 3.1. This update delivers a complete overhaul of the MCP ChatBot (mcp_chatbot.py), now rebuilt on the AG2 agent framework. As part of this redesign, we resolved limitations around supporting multiple MCP servers, delivering a more robust and true MCP solution.
In moving away from LangChain to AG2, we gain a key advantage: native support for multiple LLM families without the need for conditional code to detect which LLM is in use.
Using the TigerGraph MCP server provides uses with the ability to interact with TigerGraph through natural language processing via the ChatBot.
Example, if you have admin privileges in the dabase, you can simply type in:
run cpu & memory usage
The system will figure out which MCP sever has the tool to perform the requst.
You will recieve a formatted output like the following:
You can download the latest version at: Release TigerGraph MCP version 3.1 · custom-discoveries/TigerGraph_MCP · GitHub