mcp-agents
MCP Agents is an AI-powered browser automation tool that lets you interact with websites using natural language. Built with Streamlit, OpenAI, and Puppeteer via the Model Context Protocol (MCP), it supports multi-step navigation, interaction, and content extraction—all with simple text commands.
Documentation
🌐 MCP Agents
MCP Agents is an AI-powered interactive browser assistant built with Streamlit, OpenAI, and Puppeteer using the Model Context Protocol (MCP). Type natural language commands like *"Go to Wikipedia and search for Mars"* and the app will navigate, click, scroll, and even extract content — all hands-free.
🚀 Features
- Talk to the Web — Interact with websites using simple English commands
- Visual Automation — Take screenshots, click elements, and scroll pages effortlessly
- Flexible Agent System — Powered by the modular MCP Agent framework
- Secure Integration — API keys stored safely with optional secrets config
- Fully Interactive UI — See command results directly in the app interface
🎯 Use Cases
- 💼 Product scraping, news summarization, automated browsing workflows
- 🧪 Educational bots that walk through websites
- 📰 Daily content extraction from dynamic pages
⚙️ Requirements
- Python 3.8 or newer
- Node.js + npm (for Puppeteer server)
- OpenAI API key
⚡ Quick Start
# Clone and enter the project
$ git clone https://github.com/your-username/mcp-agents.git
$ cd mcp-agents
# Install Python dependencies
$ pip install -r requirements.txt
# Verify Node.js setup
$ node --version && npm --version
# Optional: Start Puppeteer agent
$ npx -y @modelcontextprotocol/server-puppeteer
# Run the app
$ streamlit run main.pyVisit http://localhost:8501 in your browser.
🔐 Configure Secrets
Create a file named `mcp_agent.secrets.yaml`:
openai:
api_key: "your-openai-api-key"➡️ Make sure this file is listed in `.gitignore` to avoid leaking credentials.
💬 Example Commands You Can Try
- "Go to www.wikipedia.org"
- "Search for black holes and summarize the page"
- "Click on the first heading"
- "Take a screenshot of the hero section"
- "Scroll down and extract all h2 titles"
📁 Project Overview
mcp-agents/
├── main.py # Streamlit application
├── requirements.txt # Python dependencies
├── README.md # This file
├── mcp_agent.config.yaml # Config for Puppeteer + agent setup
├── mcp_agent.secrets.yaml # API secrets (ignored from Git)
├── .gitignore # File exclusions📄 License
Licensed under the MIT License.
🙋♂️ Created By
Built with ❤️ by Harsh.
Powered by the open agent framework from MCP and inspired by LastMileAI tooling.
👨💻 Want to build your own AI-powered automation agent? Fork this repo and start customizing!
Frequently asked questions
What is mcp-agents?
mcp-agents is MCP Agents is an AI-powered browser automation tool that lets you interact with websites using natural language. Built with Streamlit, OpenAI, and Puppeteer via the Model Context Protocol (MCP), it supports multi-step navigation, interaction, and content extraction—all with simple text commands.
How do I install mcp-agents?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is mcp-agents open source?
Yes — it is hosted on GitHub at https://github.com/CodeWithHarshAI/MCP-Agents and has 2 stars.
Related MCP tools
An LLM agent that conducts deep research (local and web) on any given topic and generates a long report with citations. Built for the Model Context Protocol to
A middleware to provide an openAI compatible endpoint that can call MCP tools Python-based implementation. Trusted by 800+ developers.
Shell and coding agent on claude desktop app for the Model Context Protocol. Enhance AI assistants with powerful integrations. Python-based implementation.
Vibetest MCP - automated QA testing using Browser-Use agents Python-based implementation. Trusted by 500+ developers. Trusted by 500+ developers.
A text-based user interface (TUI) client for interacting with MCP servers using Ollama. Features include multi-server, dynamic model switching, streaming res...
基于大模型搭建的聊天机器人,同时支持 微信公众号、企业微信应用、飞书、钉钉 等接入,可选择ChatGPT/Claude/DeepSeek/文心一言/讯飞星火/通义千问/ Gemini/GLM-4/Kimi/LinkAI,能处理文本、语音和图片,访问操作系统和互联网,支持基于自有知识库进行定制企业智能客服。
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP