Multi Server Anthropic Chatbot

This chatbot helps you discover and understand research papers from arXiv using natural language and the Model Context Protocol (MCP). Hope the commented "lectures" will be useful and help clarify the stuff. :D

What it has/does

• Multi-Server Architecture: Connects to multiple MCP servers simultaneously
• Resources: Access organized paper collections by topic -- > read-only data
• Prompts: Pre-built prompt templates so the user doesn't have to do the whole prompt engineering themselves
• Tools: Servers' capabilities

Multi-Server Architecture

Connected Servers

1. Research Server (mcp_chatbot/research_server.py)

   • Tools: get_arxiv_papers, extract_info
   • Resources: papers://folders, papers://{topic}
   • Prompts: generate_search_prompt
   • Purpose: arXiv paper search and analysis

2. Fetch Server (mcp-server-fetch)

   • Purpose: Web content retrieval and processing (see "Example Queries" section to play around)

3. Filesystem Server (@modelcontextprotocol/server-filesystem)

   • Purpose: File system operations and management

Server Configuration

The system uses server_config.json to manage multiple server connections. Feel free to add your preffered ones too. :))

Start

1. (Recommended) Create and activate a virtual environment

   This keeps dependencies isolated and avoids conflicts.

   python3 -m venv .venv
   source .venv/bin/activate
   

2. Install dependencies:

   pip install -r requirements.txt
   

3. Set up your API key (create a .env file):

   ANTHROPIC_API_KEY=your_api_key_here
   

4. Start the chatbot:

   uv run mcp_chatbot.py
   

Example Queries:

• Fetch the content of this website: https://modelcontextprotocol.io/docs/concepts/architecture and save the content in the file "mcp_summary.md"
• Create a visual diagram that summarizes the content of "mcp_summary.md"

Example Workflow:

1. Search for papers: "Find papers about neural networks"
2. Browse topics: @folders to see available topics
3. Explore specific topic: @neural_networks_brain_functions
4. Get comprehensive analysis: /prompt generate_search_prompt topic=neural_networks num_papers=10

Contributing

Feel free to submit issues and enhancement requests! ☺️