This project implements a web research based agentic RAG application using CrewAI, Tavily, and Trafilatura with structured multi-agent reasoning. It includes web search, content parsing, context synthesis, and final summarization using an LLM.

• Web Search Agent using Tavily
• Content Fetcher Agent to download & parse articles with trafilatura
• Context Generator saving rich content to context_output.md
• Final Answer Agent generating summaries from parsed context
• Fully modular CrewAI agent-task pipeline

pip install -r requirements.txt

Create a .env file:

TAVILY_API_KEY=your_tavily_key
GROQ_API_KEY=your_groq_key

python3 -m agent

You’ll be prompted for a research query. The assistant will:

1. Search the web via Tavily
2. Fetch & parse top links
3. Format the content using trafilatura
4. Save it to context_output.md
5. Generate a final structured summary saved/displayed from research_output.md

├── agent.py                  # Entrypoint  
├── requirements.txt          # Dependencies
└── .env                      # Environment variables  

• gemma2-9b-it
• meta-llama/llama-4-scout-17b-16e-instruct
• llama-3.3-70b-versatile