DocQuery

Overview

DocQuery is a comprehensive project consisting of both frontend and backend components. The frontend is built with Next.js and Tailwind CSS, providing a user-friendly interface to create powerful knowledge bases for LLMs using markdown documentation. The backend leverages FastAPI to provide endpoints for different agents built using the langgraph library and integrated into the FastAPI application using CopilotKitSDK.

Video: Click to view

Setup

Frontend

1. Clone the repository:

   git clone https://github.com/md-abid-hussain/docquery.git
   cd docquery-frontend

2. Install dependencies:

   npm install

3. Copy the example environment file and fill in the required values:

   cp .env.example .env

4. Start the development server:

   npm run dev

5. The application will be available at http://localhost:3000.

Backend

1. Clone the repository:

   git clone <repository-url>
   cd docquery-backend

2. Create and activate a virtual environment:

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

3. Install dependencies:

   pip install -r requirements.txt

4. Copy the example environment file and fill in the required values:

   cp .env.example .env

5. Start the FastAPI server:

   uvicorn app.main:app --reload

6. The application will be available at http://127.0.0.1:8000.

Features

Frontend

• GitHub Integration: Easily fetch repository details and select markdown files for ingestion.
• Markdown Processing: Ingest and process markdown files used for documentation or detailed explanations.
• Knowledge Base Creation: Build comprehensive knowledge bases for LLMs from your documentation.
• AI-Powered Insights: Leverage advanced LLMs to generate insights and answer queries based on your knowledge base.

Backend

Ingestion Agent

The Ingestion Agent is defined in agent.py. It consists of the following nodes:

• Ingestion Node
• Verify Ingestion Node

QA Agent

The QA Agent is defined in agent.py. It consists of the following nodes:

• Retrieve Node
• Chat Node

API Endpoints

• GET /: Returns a simple greeting message.
• POST /copilotkit: Endpoint for interacting with the agents.

Environment Variables

The application requires the following environment variables to be set:

Frontend

• TOGETHER_AI_API_KEY
• DATABASE_URL

Backend

• GOOGLE_API_KEY
• TOGETHER_API_KEY
• GITHUB_PERSONAL_ACCESS_TOKEN
• MONGODB_ATLAS_CLUSTER_URI

These can be set in the respective .env files.

License

This project is licensed under the MIT License.