🔬 ResearchMind — Multi-Agent AI Research System

Four specialized AI agents that collaborate to search, read, write, and critique — producing polished research reports on any topic.

---

📖 Overview

ResearchMind is a production-ready multi-agent AI pipeline that automates the entire research workflow. You provide a topic; four autonomous agents handle the rest:

Agent	Role
🔍 Search Agent	Queries the web via Tavily for recent, reliable information
📄 Reader Agent	Picks the best URL and scrapes deep content from it
✍️ Writer Chain	Synthesizes all gathered research into a structured report
🧐 Critic Chain	Reviews the report and provides a score + constructive feedback

The project ships with a polished Streamlit web UI that visualizes every pipeline step in real time, plus a headless CLI entrypoint (pipeline.py).

---

🏗️ Architecture

┌─────────────────────────────────────────────────────┐
│                   User (UI / CLI)                   │
└────────────────────────┬────────────────────────────┘
                         │ topic
                         ▼
              ┌──────────────────────┐
              │   Search Agent       │  ← Tavily Web Search
              │  (LangChain Agent)   │
              └──────────┬───────────┘
                         │ search_results
                         ▼
              ┌──────────────────────┐
              │   Reader Agent       │  ← BeautifulSoup Scraper
              │  (LangChain Agent)   │
              └──────────┬───────────┘
                         │ scraped_content
                         ▼
              ┌──────────────────────┐
              │   Writer Chain       │  ← GPT-4o-mini (LLM)
              │  (LCEL Chain)        │
              └──────────┬───────────┘
                         │ report
                         ▼
              ┌──────────────────────┐
              │   Critic Chain       │  ← GPT-4o-mini (LLM)
              │  (LCEL Chain)        │
              └──────────┬───────────┘
                         │ feedback
                         ▼
              ┌──────────────────────┐
              │   Final Output       │
              │  (Report + Review)   │
              └──────────────────────┘

---

✨ Features

• 🤖 Multi-agent orchestration — independent agents with distinct responsibilities
• 🌐 Real-time web search via Tavily (not just static knowledge)
• 🕸️ Intelligent web scraping with BeautifulSoup content extraction
• 📝 Structured report generation (Introduction → Key Findings → Conclusion → Sources)
• 🧪 Automated quality critique with scoring (X/10) and improvement suggestions
• 💻 Dual interface — Beautiful Streamlit UI and CLI mode
• ⬇️ Download reports as .md files directly from the UI
• 🎨 Dark-mode premium UI with animated pipeline status cards

---

🚀 Getting Started

Prerequisites

• Python 3.10+
• An OpenAI API key (get one here)
• A Tavily API key (get one here — free tier available)

1. Clone the repository

git clone https://github.com/<your-username>/multi-research-agent.git
cd multi-research-agent

2. Create and activate a virtual environment

python -m venv venv
source venv/bin/activate      # Linux / macOS
# venv\Scripts\activate       # Windows

3. Install dependencies

pip install -r requirements.txt

4. Configure environment variables

cp .env.example .env

Open .env and fill in your API keys:

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

⚠️ Never commit your .env file. It is already listed in .gitignore.

5. Run the Streamlit app

streamlit run app.py

Open your browser at http://localhost:8501.

6. (Alternative) Run via CLI

python pipeline.py

You will be prompted to enter a research topic, and the full pipeline will run in your terminal.

---

📁 Project Structure

multi-research-agent/
├── app.py              # Streamlit web application (main UI)
├── pipeline.py         # Headless CLI entrypoint
├── agents.py           # Agent & chain definitions (Search, Reader, Writer, Critic)
├── tools.py            # LangChain tools (web_search, scrape_url)
├── requirements.txt    # Python dependencies
├── .env.example        # Environment variable template (safe to commit)
├── .env                # Your actual secrets (NEVER commit this)
├── .gitignore
├── LICENSE
└── README.md

---

🔧 Configuration

Variable	Description	Required
OPENAI_API_KEY	Your OpenAI API key	✅ Yes
TAVILY_API_KEY	Your Tavily Search API key	✅ Yes

You can modify the model used in agents.py:

llm = ChatOpenAI(model="gpt-4o-mini", temperature=0)
# Change to "gpt-4o" for higher quality, or "gpt-3.5-turbo" for lower cost

---

🧩 How the Agents Work

🔍 Search Agent

Uses the web_search tool (backed by Tavily) to retrieve up to 5 recent, reliable web results for the given topic — including title, URL, and a content snippet.

📄 Reader Agent

Takes the search results, picks the most relevant URL, and scrapes its full text using requests + BeautifulSoup. Scripts, styles, navigation, and footers are stripped for clean content.

✍️ Writer Chain

An LCEL chain that feeds the combined search + scraped data into GPT-4o-mini with a structured prompt. Outputs a full markdown report with:

• Introduction
• Key Findings (≥3 points)
• Conclusion
• Sources

🧐 Critic Chain

An LCEL chain that reviews the generated report and returns:

• Score: X/10
• Strengths: bullet points
• Areas to Improve: bullet points
• One-line verdict

---

🤝 Contributing

Contributions are welcome! Please read CONTRIBUTING.md first.

1. Fork the repository
2. Create your feature branch: git checkout -b feature/amazing-feature
3. Commit your changes: git commit -m 'feat: add amazing feature'
4. Push to the branch: git push origin feature/amazing-feature
5. Open a Pull Request

---

📄 License

This project is licensed under the MIT License — see the LICENSE file for details.

---

🙏 Acknowledgements

• LangChain — Agent & chain orchestration
• Streamlit — Web UI framework
• Tavily — AI-optimized web search API
• OpenAI — GPT-4o-mini language model
• BeautifulSoup — Web scraping

---

_{Built with ❤️ · Powered by LangChain multi-agent pipeline}