AI Research Agent with LangGraph

An intelligent research assistant that uses LangGraph to clarify research queries, fetch relevant papers from Google Scholar, and save results to markdown files.

Features

• Query Clarification: Uses OpenAI with structured output parsing to refine and structure research queries
• Google Scholar Integration: Fetches relevant research papers using the Scholarly library
• LangGraph Workflow: Implements a multi-step agent workflow with state management
• FastAPI Server: Provides a REST API for easy interaction
• Markdown Output: Saves research results with paper names and links to markdown files

Setup

1. Install Dependencies

pip install -r requirements.txt

2. Configure Environment Variables

Choose your LLM provider (OpenAI, DeepSeek, OpenRouter, or Gemini):

cp .env.example .env
# Edit .env and set:
# - LLM_PROVIDER=openai, deepseek, openrouter, or gemini
# - Add the corresponding API key

For OpenAI:

LLM_PROVIDER=openai
OPENAI_API_KEY=sk-your-key-here

For DeepSeek:

LLM_PROVIDER=deepseek
DEEPSEEK_API_KEY=your-key-here

For OpenRouter:

LLM_PROVIDER=openrouter
OPENROUTER_API_KEY=sk-or-your-key-here

For Gemini:

LLM_PROVIDER=gemini
GOOGLE_API_KEY=your-google-api-key-here

3. Run the Agent

Option 1: Direct Python Script

python agent.py "your research query here"

Option 2: FastAPI Server

python api_server.py

Then make a POST request:

curl -X POST "http://localhost:8000/research" \
  -H "Content-Type: application/json" \
  -d '{"query": "machine learning for climate change"}'

Project Structure

• agent.py - Main LangGraph agent implementation
• query_clarifier.py - OpenAI-based query clarification with structured output
• scholar_fetcher.py - Google Scholar API integration
• api_server.py - FastAPI server for the research agent
• models.py - Pydantic models for structured data
• results/ - Directory where markdown results are saved

How It Works

1. Query Input: User provides a research query
2. Clarification: Agent uses OpenAI to clarify and structure the query
3. Paper Fetching: Searches Google Scholar for relevant papers
4. Result Writing: Saves paper names and links to a markdown file

Example Output

The agent creates markdown files in the results/ directory with content like:

# Research Results: PerFed Personalized Fedeated Learning

**Clarified Query**: Personalized Federated Learning, PerFed (Personalized Federated), Federated Learning Personalization, Client-Specific Model Adaptation

## Papers Found

1. [PerFED-GAN: Personalized federated learning via generative adversarial networks](https://ieeexplore.ieee.org/abstract/document/9766407/)
2. [Perfedmask: Personalized federated learning with optimized masking vectors](https://openreview.net/forum?id=hxEIgUXLFF)
...

Requirements

• Python 3.9+
• API key for one of: OpenAI, DeepSeek, OpenRouter, or Gemini
• Scholarly - for Google Scholar Feed Crawlng