Deep-Source Research Agent

A full-stack AI research agent built with Python, Streamlit, and Google's Gemini. It autonomously researches complex topics by browsing the web, cross-referencing sources, auditing for bias, and generating a cited Markdown report with confidence scores.

Features

1. Plan: Generates distinct search queries based on your topic.
2. Search & Scrape: Asynchronously searches the web using DuckDuckGo and extracts full body text from top results via Trafilatura.
3. Audit: Analyzes findings using Gemini to detect contradictions, missing context, and biased language.
4. Synthesize: Synthesizes the audited information into a final, highly structured markdown report complete with inline citations and section-based Confidence Scores.

Prerequisites

• Python 3.9+
• A Google Gemini API Key

Installation

1. Clone the repository:

   git clone <your-repo-url>
   cd deep-source-research-agent

2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install the dependencies:

   pip install -r requirements.txt

4. Environment Variables: Rename .env.example to .env and add your Google Gemini API key:

   GEMINI_API_KEY=your_gemini_api_key_here

Usage

Start the Streamlit application:

streamlit run app.py

The application will open in your default browser at http://localhost:8501. Enter a complex research topic and click "Start Research" to watch the agent work!

Tech Stack

• Frontend: Streamlit
• LLM/Brain: Google Generative AI (Gemini Flash)
• Web Search: duckduckgo-search
• Web Scraping: Trafilatura