A smart multi-tool AI research assistant that reads PDFs, searches the web, queries Wikipedia, solves math, and explains every step it takes — all using LangChain Agents and Mistral-7B via OpenAI.

✅ Built with: LangChain, Streamlit, Mistral-7B, Tavily, PyMuPDF, Wikipedia, PythonREPL

• 📄 Summarize PDFs
• 🌐 Search the Web using Tavily
• 📚 Get facts from Wikipedia
• 🧮 Solve math with Python REPL
• 🔁 Custom reverse tool
• 💬 Multi-turn memory
• 🪄 Shows Chain-of-Thought (Thought → Action → Result)
• ✅ Tool usage checklist
• 🎨 Clean Streamlit UI

• Python 3.9+
• API Keys from:
  • OpenAI
  • Tavily

# 1. Clone this repo
git clone https://github.com/Rakshath66/Assistant-Research-Agent-LangChain.git
cd Assistant-Research-Agent-LangChain

# 2. Create virtual environment (optional but recommended)
python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows

# 3. Install dependencies
pip install -r requirements.txt

# 4. Add your API keys in .env
echo "OPENAI_API_KEY=your_openai_key" >> .env
echo "TAVILY_API_KEY=your_tavily_key" >> .env

# 5. Launch the app
streamlit run src/streamlit_app.py

Assistant-Research-Agent-LangChain/
├── src/
│   ├── streamlit_app.py     # Main UI and agent setup
│   └── agent_tools.py       # Custom LangChain tools
├── requirements.txt         # All Python dependencies
├── .env.example             # Sample env format
└── README.md

Create a .env file (or set in Streamlit Secrets):

OPENAI_API_KEY=your_openai_key
TAVILY_API_KEY=your_tavily_key

If you're using Streamlit Cloud, paste this into Settings → Secrets:

OPENAI_API_KEY = "sk-..."
TAVILY_API_KEY = "tvly-..."

• “What projects did Rakshath work on in the uploaded PDF?”
• “Summarize Wikipedia on LangChain.”
• “Who is Elon Musk? Square the number of years since his birth.”
• “Reverse the string ‘LangChain Rocks!’”

We welcome contributions! Here's how you can help:

1. Fork this repository
2. Clone your fork: git clone https://github.com/Rakshath66/Assistant-Research-Agent-LangChain.git
3. Create a new branch: git checkout -b feature/my-feature
4. Make your changes, commit, and push: git commit -m "Add: my feature" git push origin feature/my-feature
5. Open a Pull Request with a description of your changes

🔍 Please write clean readable code, add docstrings if needed, and test your features!

MIT License © Rakshath U Shetty

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software... [rest of MIT license]

It helps others discover it, and motivates me to build more free AI tools. Also, feel free to open issues, request features, or contribute.

• Open source forever
• Designed for learning, research, and practical use
• Reach out via issues or PRs — ideas welcome!

What Next? This project is just the beginning. We’re building a production-ready RAG Agent that evolves over time. Here's the full roadmap after completing the basic agent.

• 🧠 Chain-of-Thought Agent using LangChain
• 🔧 Tools: PDF summarization, Web search (Tavily), Wikipedia, Calculator
• 🧮 Built-in Memory
• 💡 Reasoning trace (Thought → Action → Result)
• 🎨 Streamlit UI with tool checklist

Let the agent auto-chain steps like:

“Summarize top 3 news articles about XYZ”
⟶ Search → Pick top links → Read + Summarize → Combine insights

• summarize_url — Read and summarize a web page
• multi_url_summary — Loop over list of URLs and extract key info
• web_search_chain — Combined search + read + summarize

Let the agent remember facts about the user or previous answers.

Example:

“Who is the CEO of Tesla?”
→ Even tomorrow, it remembers your earlier query.

🧰 Tools & Techniques:

• LangChain’s VectorStoreRetrieverMemory
• Store structured Q&A facts or timelines
• Indexed long-term memory support

Enhance document summarization:

• 🔍 Show exactly which chunks from the PDF were used in the answer
• 📄 Highlight matching text in context
• 🗂️ Support multi-PDF ingestion with separate memory per doc

Turn this logic into a production API backend:

• 🔌 Framework: FastAPI or Litestar
• 🧠 Endpoint: /ask → returns final answer + trace
• 💬 Useful for building your own chatbot or research SaaS

Make the agent smarter at real-world decisions:

• 🌐 Use LangGraph or LCEL to handle branching logic
• 🤖 Tools like:
  • YouTubeTranscriptTool
  • Google Scholar Reader
  • News Aggregator
• 🧠 Agent decides:
  • Should I search?
  • Should I summarize?
  • Can I skip?

⭐ Want to contribute to any phase? Fork the repo, build a feature, and submit a PR! The roadmap is open for collaboration.

Let me know if you want:
- `LICENSE` file (MIT version)
- A matching `.env.example` file
- `demo/screenshot.png` placeholder
- `contributing.md` file

All of this helps boost your open-source visibility!