# AI agent

An AI-driven browser automation server based on the MCP protocol, supporting natural language control of web page operations and in-depth web research.

The Graphiti MCP Server is a server based on multi - project knowledge graphs, supporting the rapid construction and management of temporary knowledge graphs for AI agents to query through the model context protocol. This project extends the official getzep/graphiti MCP Server, adds support for multiple servers and a single database, and provides a CLI tool with an optimized developer experience.

This project creates an AI agent based on the n8n platform that can discover and use Model Context Protocol (MCP) server tools through the Nostr network. The agent searches for available tools by querying Data Vending Machines (DVM) in the network, sends requests, and obtains responses, thereby expanding the functionality of the LLM to use MCP tools that are not installed.

This project is an interaction server for the Open eClass platform based on the MCP protocol, supporting the SSO authentication system of the University of Athens. It provides functions such as course management and session maintenance, enabling AI agents to securely interact with the eClass platform.

The LINE Bot MCP Server is a server that implements the Model Context Protocol (MCP), used to connect an AI agent to a LINE official account and supports sending text and flexibly laid-out messages.

An MCP server implementation for AdGuard Home, supporting querying and managing DNS records, filtering rules, etc. through an AI agent.

mcp - agent is a lightweight AI agent framework based on the Model Context Protocol (MCP), which supports building efficient agent applications through a simple and composable model. It integrates the lifecycle management of the MCP server and implements multiple workflow models, such as parallel processing, routing, evaluation optimization, etc., and is suitable for various AI application scenarios such as multi - agent collaboration and human - machine interaction.

This project demonstrates how to integrate OpenAI, Azure OpenAI, and GitHub models through the MCP protocol to build a terminal application for a TODO list agent. The project includes three major components: MCP host, client, and server, supports HTTP and SSE protocols, and provides tool invocation and database storage functions.

The ThingsBoard MCP Server is a service that provides a natural language interface for LLMs and AI agents, allowing interaction with the ThingsBoard IoT platform through natural language commands to achieve functions such as device management, data query, and analysis.

Sequential Thinking MCP is an MCP server that supports AI agents in advanced metacognition and dynamic reflective problem-solving, guiding the step-by-step execution of complex tasks through virtual thought recording and planning.

Project exploring MCP servers and AI agents

A Gmail email management server based on the Model Context Protocol (MCP), supporting searching, reading, deleting, and sending emails through an AI agent. It needs to be used with a voice interaction client.

The Azure MCP Server is a server that implements the MCP protocol, providing seamless connection between AI agents and Azure services (such as Storage and Cosmos DB), and supporting the conversion of natural language commands into Azure operations.

PG-MCP is a model context protocol server designed for PostgreSQL databases, providing APIs for AI agents to discover, connect, query, and understand databases. It supports multi - database connection, rich catalog information, and query explanation functions.

The TMF620 MCP Server is a middleware that connects AI agents with the remote TMF620 product catalog management API, providing functions for querying and creating catalogs, product offerings, and product specifications.

Discourse MCP is a Model Context Protocol server that exposes Discourse forum functions as tools to AI agents. It supports operations such as searching, reading posts, and creating posts, provides read-only and writable modes, and supports remote tool discovery and dynamic registration.

An OpenAI agent server based on the MCP protocol, providing various professional agents (such as web search, file search, and computer operation) and a multi-agent coordinator, which can interact with clients (such as the Claude desktop application) through the MCP protocol.

automcp is a tool that can easily convert existing AI agent frameworks (such as CrewAI, LangGraph, etc.) into MCP servers to achieve standardized interface access. It supports the adaptation and deployment of multiple frameworks and provides functions for quick startup and configuration.

iFlytek Workflow MCP Server is an AI workflow scheduling system based on the MCP protocol, supporting multiple node types and advanced orchestration modes, enabling intelligent business process automation.

A Minecraft model control protocol server based on Mineflayer, providing a standardized JSON - RPC interface for AI agents to control Minecraft robots.

MCP speech synthesis server based on Edge-TTS

PG-MCP is a model context protocol server designed for PostgreSQL databases. It provides APIs for AI agents to connect to, query, and understand databases, and supports multiple database connections, rich metadata extraction, and query analysis functions.

A high-performance PostgreSQL MCP server based on Bun and TypeScript, providing multi-database interaction capabilities for AI agents.

This is an MCP server project built for the Norwegian power supplier Tibber, allowing AI agents to access the Tibber API through the Model Context Protocol to query energy prices and electricity consumption data. The project provides a local running solution and supports connection through clients such as Claude Desktop.

The Agentic MCP Client is an independent AI agent runner that interacts with models such as Claude, BedRock, and OpenAI through MCP protocol tools, supporting autonomous cloud operation and secure system interaction.

A Django application that implements a Firebase Model Context Protocol (MCP) server, enabling AI agents to interact with Firebase services through a standardized protocol. It provides 14 Firebase tools covering authentication, database, and storage functions, supporting quick testing and full integration.

The Screeny MCP Server is a privacy-first screenshot service designed specifically for macOS, allowing AI agents to securely capture screenshots of pre-approved application windows and providing a visual context for development and debugging tasks.

An MCP server based on The Graph that provides indexed blockchain data query services for AI agents, supporting subgraph structure retrieval and GraphQL query execution.

This project implements an adapter between Langchain and the Box MCP server, integrating AI capabilities through tools and agents, supporting dynamic tool loading, creating React-style agents, and providing a friendly console interaction interface.

MCP BatchIt is a batching server for MCP tool calls. By aggregating multiple MCP tool calls into one request, it significantly reduces the communication overhead and token consumption of AI agents.