No description available
A powerful web research pipeline MCP built using the Atomic Agents framework.
For a full breakdown of the code, please check out this article
For now, this requires an API key for Tavily and also an OpenAI API key . In the future, I plan to make this more configurable so you can use SearxNG instead of Tavily or use Groq or Anthropic instead of OpenAI, thanks to the Atomic Agents framework this is all easy and possible, it just requires a bit of time and I wanted to get the initial project out there. Feel free to contribute, though!
Do you like this project? Please consider a small donation, it means the world to me!
This project implements an advanced web research pipeline that leverages the Model Context Protocol (MCP) and Atomic Agents to provide comprehensive answers to research questions. The pipeline automates the entire research process:
The system follows a modular architecture based on the MCP client-server model:
User Question โ Query Generation โ Web Search โ Content Scraping โ Answer Synthesis โ Formatted Response
Clone the repository:
git clone https://github.com/yourusername/agentic-research-mcp.git
cd agentic-research-mcp
Install dependencies:
pip install -e .
Set up environment variables:
# Create a .env file in the project root
echo "TAVILY_API_KEY=your_tavily_api_key" > .env
echo "OPENAI_API_KEY=your_openai_api_key" >> .env
Start the MCP server:
python -m atomic_research_mcp.server
OR
[rootfolder]\atomic-reseearch-mcp\.venv\Scripts\atomic-research
OR configure it in, for example, Cursor
Run the test client to verify functionality:
python test_client.py
The system returns comprehensive research results including:
atomic_research_mcp/
โโโ agents/
โ โโโ query_agent.py # Generates optimized search queries
โ โโโ qa_agent.py # Synthesizes answers from scraped content
โโโ tools/
โ โโโ tavily_search.py # Interface to Tavily search API
โ โโโ webpage_scraper.py # Extracts and processes web content
โโโ server.py # MCP server implementation
โโโ config.py # Configuration settings
The system can be configured through environment variables:
TAVILY_API_KEY
: Required for web search functionalityOPENAI_API_KEY
: Required for AI agent operationsOPENAI_MODEL
: Optional, defaults to "gpt-4o-mini"Contributions are welcome! Please feel free to submit a Pull Request.
{ "mcpServers": { "atomic-research-mcp": { "command": "python", "args": [ "-m", "atomic_research_mcp.server" ] } } }
Related projects feature coming soon
Will recommend related projects based on sub-categories