LangGraph-GUI

LangGraph-GUI is a user-friendly interface for managing and visualizing Node-Edge workflows with LangGraph. It supports creating, editing, and running workflows locally using language models by Ollama .

This repo is implemented with Qt. If you want ReactFlow version frontend, see LangGraph-GUI-ReactFlow

This is node-edge based gui will export to json as saved graph. And run the graph by LangGraph.

If you want to learn more about LangGraph, we have LangGraph for dummy : LangGraph-learn

Explanation

Start Node

A graph(json) only have one start node, this will mapping to LangGraph START

Step Node

Step Node will mapping to LangGraph add_node, you can drag edge from left node(right port) to right node(left port)

if you drag two node toward to each other, can create cycle.

Step Node with tool

Step node fill tool will call function, the tool definition need a tool node

Tool Node

Tool node need write real python function code, and need @tool decorator like LangChain Custom Tools

Contition Node

CONDITION node will mapping to LangGraph conditional_edge

• green edge is true case path
• red edge is false case path

Environment Setup

Front-End GUI

To install the required dependencies for the front-end GUI, run:

pip install PySide6

Back-End

To install the required dependencies for the LangGraph, run:

pip install langchain langchain-community langchain-core langgraph

Running the Application

Front-End

To start the front-end GUI, execute:

python frontend.py

This will allow you to read and write JSON files representing DAG workflows for CrewAI.

Back-End

If want to run local llm, need run Ollama first

ollama serve

Then run the back-end locally with a model such mistral, use:

python backend.py --graph example.json --llm gemma2 --tee output.log

This command will parse the specified JSON file into Graph.

Building the Application

Front-End GUI

To build the front-end GUI into a standalone executable, follow these steps:

1. Install PyInstaller:

   pip install pyinstaller

2. Navigate to the source directory:

   cd src

3. Run PyInstaller with the necessary hooks:

   pyinstaller --onefile --additional-hooks-dir=. frontend.py