mcp-agent-graph

 English | [中文](README_CN.md) 📚 [Documentation](https://keta1930.github.io/mcp-agent-graph/#) | 📦 [PyPI Package](https://pypi.org/project/mcp-agent-graph/) ## Table of Contents 1. [Roadmap](#1-roadmap) 2. [Deployment Guide](#2-deployment-guide) - [Clone Project](#21-clone-project) - [Start Docker Services](#22-start-docker-services) - [Start Frontend Service](#23-start-frontend-service) - [Backend Deployment](#24-backend-deployment) 3. [Core Features](#3-core-features) 4. [Frontend Feature Showcase](#4-frontend-feature-showcase) - [deepresearch (Agent Generated)](#41-deepresearch-agent-generated) - [corporate_ethics_dilemma_v2 (Agent Generated)](#42-corporate_ethics_dilemma_v2-agent-generated) 5. [Citation](#5-citation) 6. [WeChat Group](#6-wechat-group) 7. [Milestone](#7-milestone) 8. [Star History](#8-star-history) ## 1. Roadmap [View the Roadmap](assets/roadmap_en.md) ## 2. Deployment Guide ### 2.1. Clone Project ```bash git clone https://github.com/keta1930/mcp-agent-graph.git ``` ### 2.2. Start Docker Services ```bash # Copy environment configuration file, or use .env file directly cd docker/mag_services cp .env.example .env # Start Docker services docker-compose up -d ``` **Default .env Service Addresses:** - MongoDB Express (Database Management): http://localhost:8081 - MinIO Console (File Storage): http://localhost:9011 ### 2.3. Start Frontend Service ```bash # Enter frontend directory cd frontend # Install dependencies and start npm install npm run dev ``` **Access Address:** http://localhost:5173 ### 2.4. Backend Deployment **Option 1: PyPI Installation (Recommended)** ```bash pip install mcp-agent-graph >>> mag.start() ``` **Option 2: Source Code Deployment** ```bash git clone https://github.com/keta1930/mcp-agent-graph.git cd mcp-agent-graph # Using uv (Recommended) uv sync cd mag nohup uv run python main.py > app.log 2>&1 & # Or using pip pip install -r requirements.txt cd mag nohup python main.py > app.log 2>&1 & ``` **Service Addresses:** - Backend API: http://localhost:9999 - MCP Client: http://localhost:8765 ## 3. Core Features #### 3.1. System-Level Agent System-level Agent helps users customize Agent Workflow/Agent Graph and MCP tools #### 3.2. Visual Graph Editor Frontend creation of intelligent agent workflows, what you see is what you get #### 3.3. Graph Nesting Agent reusability, any graph can be used as a node in other graphs, building hierarchical intelligent systems #### 3.4. Task Scheduling System Support for timed and periodic execution of Agents, batch concurrent processing #### 3.5. Graph to MCP Service One-click export of agents as standard MCP services, callable by Claude, Cline, etc. #### 3.6. Agent Trading and Transfer Complete agent packaging, sharing, and deployment solution #### 3.7. Python SDK Deep Integration `pip install mcp-agent-graph` to build Agents using Python #### 3.8. Prompt Registration Management One-stop prompt management, register and reuse prompt templates ## 4. Frontend Feature Showcase ### 4.1. deepresearch (Agent Generated) #### Deep analysis of user questions, multi-round intelligent retrieval, and comprehensive research system that generates visualized HTML web pages  --- ### 4.2. corporate_ethics_dilemma_v2 (Agent Generated) #### AI CFO Alex faces complex corporate ethical choices, exploring AI decision-making mechanisms in conflicts of interest  --- ## 5. Citation If you find MCP Agent Graph helpful for your research or work, please consider citing it: ```bibtex @misc{McpAgentGraph, author = {Yixin Yan and mcp-agent-graph Contributors}, title = {mcp-agent-graph: A Multi-agent Collaboration Framework}, year = {2025}, howpublished = {\url{https://github.com/keta1930/mcp-agent-graph}}, note = {GitHub repository. Corresponding author: Yixin Yan}, urldate = {2025-04-24} } ``` ## 6. WeChat Group  ## 7. Milestone  ## 8. Star History [](https://www.star-history.com/#keta1930-mcp-agent-graph&Date)