LangGraph Nexus

A production-style AI assistant built with LangGraph, MCP, RAG, and 8+ tools — all running in a sleek Streamlit interface.

Live Demo

Coming soon — deployment in progress

What Is This?

This is not your average chatbot.

It's a fully agentic AI assistant that can:

• Chat with your uploaded PDF documents
• Track, update, and analyze your expenses via a local MCP server
• Search the live web for current information
• Fetch real-time stock prices and trends
• Get live weather for any city
• Search ArXiv research papers
• Look up anything on Wikipedia
• Perform calculations
• Remember conversations across sessions using SQLite checkpointing

Architecture

User
 │
 ▼
Streamlit Frontend  (streamlit_frontend.py)
 │
 ▼
┌─────────────── LangGraph Agent ───────────────────────┐
│                                                        │
│   chat_node  ◄──────────────►  tool_node              │
│   (LLM decides)                (executes tools)        │
│        │                            │                  │
│        ▼                            ▼                  │
│   GPT-4o-mini                  All 15 tools            │
│   (via OpenRouter)                                     │
│        │                                               │
│        ▼                                               │
│   SQLite Checkpointer  (chatbot.db — conversation      │
│   AsyncSqliteSaver      memory across sessions)        │
└────────────────────────────────────────────────────────┘
 │                                │
 ▼                                ▼
General Tools              MCP Tools (stdio)
- RAG (FAISS + PDF)        FastMCP Server → expenses.db
- DuckDuckGo search          add / list / update /
- Wikipedia                  delete / summarize
- Calculator
                                  │
External API Tools                ▼
- Stock price/history        Dashboard.py
- Live weather               (Matplotlib analytics)
- ArXiv papers

Tech Stack

Project Structure

 LangGraph_nexus/
 ┣-- langgraph_backend.py   # Agent, tools, graph, checkpointer
 ┣-- streamlit_frontend.py  # Streamlit chat UI
 ┣-- dashboard.py           # Expense analytics dashboard
 ┣-- Dockerfile
 ┣-- docker-compose.yml
 ┣-- .dockerignore
 ┣-- .env.example           # Template for environment variables
 ┣-- requirements.txt
 ┗-- .env                   # Your API keys (not committed)

 MCP_files/
 ┗-- main.py                # Local MCP ExpenseTracker server
 ┗-- categories.json

Tools Available

Setup & Installation

1. Clone the repository

git clone https://github.com/DataDoodle14/LangGraph_Nexus.git
cd LangGraph_Nexus

2. Create and activate a virtual environment

conda create -n nexus_env python=3.11
conda activate nexus_env

3. Install dependencies

pip install -r requirements.txt

4. Set up your .env file

# Mac/Linux
cp .env.example .env

# Windows
copy .env.example .env

Then open .env and fill in your API key:

OPENAI_API_KEY=your_openrouter_api_key_here
OPENAI_BASE_URL=https://openrouter.ai/api/v1

Get a free API key at openrouter.ai

5. Start the chatbot

streamlit run streamlit_frontend.py

6. Start the analytics dashboard (optional, separate tab)

streamlit run dashboard.py

Docker Setup

Run the entire app (chatbot + dashboard) with a single command — no Python or pip install needed, just Docker.

Prerequisites

• Docker Desktop installed and running

1. Clone the repository

git clone https://github.com/DataDoodle14/LangGraph_Nexus.git
cd LangGraph_Nexus

2. Set up your .env file

# Mac/Linux
cp .env.example .env

# Windows
copy .env.example .env

Then open .env and fill in your API key:

OPENAI_API_KEY=your_openrouter_api_key_here
OPENAI_BASE_URL=https://openrouter.ai/api/v1

Get a free API key at openrouter.ai

3. Build and start all services

docker-compose up --build

4. Access the apps

5. Stop all services

docker-compose down

6. Stop and remove all data (full reset)

docker-compose down -v

The -v flag deletes the shared Docker volume, wiping both chatbot.db and expenses.db.

Notes

• Both services share a Docker volume (shared_data) mounted at /tmp, so expense data written by the chatbot is instantly visible in the dashboard.
• Your .env file is never copied into the image — it is injected at runtime via env_file.
• The nexus_env/ virtual environment and all *.db files are excluded from the image via .dockerignore.

Expense Dashboard

A separate Streamlit dashboard that visualizes your expense data:

• KPI cards — Total Spent, No. of Expenses, Top Category, Average
• Monthly spending trend (line chart)
• Category breakdown (pie + bar chart)
• Per-category trend over months
• Filterable raw data table

Reset / Clear Data

Clear chat history:

# Local
del chatbot.db

# Docker (resets shared volume entirely)
docker-compose down -v

Clear expense data (local):

python -c "import tempfile; print(tempfile.gettempdir())"
# Then delete expenses.db from that path
del "C:\Users\admin\AppData\Local\Temp\expenses.db"

Screenshots

Homepage Tools:

• Weather & stock price
• Basic Calculator & DuckDuckGo Search
• ArXiv research papers
• RAG tool
• MCP Local server : Expense

Expense Dashboard

Requirements

See requirements.txt for the full pinned dependency list.

Author

Made with by DataDoodle14

License

MIT License — feel free to use, modify, and build on top of this!