[HAI5014] Integrating lightRAG into Open WebUI

Posted Jun 5, 2025

Kingos connecting a knowledge base to their chatbot (by ChatGPT)

By Pim Kamphuis

5 min read

[HAI5014] Integrating lightRAG into Open WebUI

Disclaimer: This blog provides instructions and resources for the workshop part of my lectures. It is not a replacement for attending class; it may not include some critical steps and the foundational background of the techniques and methodologies used. The information may become outdated over time as I do not update the instructions after class.

Install lightRAG

As discussed in class, today we’ll explore the lightRAG package, which allows us to feed external knowledge into our Open WebUI chatbot.

Update the requirements.txt

To ensure all required Python packages (including lightRAG) are installed, add the following lines to your requirements.txt file:

beautifulsoup4
ipykernel
lightrag-hku
markdownify
stealth-requests
tqdm

Install the python packages

To install the necessary packages, run the following command in your development container terminal:

pip install -r requirements.txt

or, if you used the UV installation method:

uv pip install -r requirements.txt

Add environment variables for lightRAG

Before running lightRAG, you’ll need to configure several environment variables. Therefore, open your .env file in the root of your Open WebUI project and add the following lines:

## Environment variables for LightRAG server

# LightRAG Storage Configuration
LIGHTRAG_KV_STORAGE=PGKVStorage
LIGHTRAG_VECTOR_STORAGE=PGVectorStorage
#LIGHTRAG_GRAPH_STORAGE=PGGraphStorage
LIGHTRAG_DOC_STATUS_STORAGE=PGDocStatusStorage

# PostgreSQL Configuration (from Supabase string above)
POSTGRES_USER=supabase_database_user
POSTGRES_PASSWORD=supabase_database_password
POSTGRES_HOST=supabase_database_host
POSTGRES_PORT=5432
POSTGRES_DATABASE=postgres
POSTGRES_MAX_CONNECTIONS=12

# Azure OpenAI API Configuration (find at Azure OpenAI Studio)
LLM_BINDING=azure_openai
AZURE_OPENAI_ENDPOINT=https://<your-custom-endpoint>.openai.azure.com
AZURE_OPENAI_API_KEY=<your-azure-openai-api-key>
AZURE_OPENAI_API_VERSION=2025-01-01-preview

LLM_MODEL=gpt-4o-mini
EMBEDDING_BINDING=azure_openai
EMBEDDING_MODEL=text-embedding-3-small
EMBEDDING_DIM=1536

You can find all the information for the postgreSQL configuration as fragments in your Supabase database URL. For the Azure OpenAI configuration, you can find the necessary information in the deployments section of your Azure OpenAI Studio.

In the example above, LIGHTRAG_GRAPH_STORAGE is commented out because Supabase does not support graph storage. Therefore, lightRAG will store the graph locally on the disk. If you want to use cloud storage for your graph, you can have a look at neo4j or other graph databases.

Explore lightRAG server

The lightRAG server offers a Web UI for document indexing, knowledge graph exploration, and a basic RAG query interface. After installing lightRAG and setting the environment variables, you can start the lightRAG server by running:

dotenv run lightrag-server

This will start the lightRAG server

Check the lightRAG server web interface via http://localhost:9621
Upload a sample document to verify your configuration and observe how lightRAG processes it
Delete the document again, so that you can start with a fresh library for the next steps
Check out the lightRAG API via http://localhost:9621/docs to see the available endpoints and their usage

Automatic document feeding using lightRAG API

In this section, we will explore how to automatically feed documents into the lightRAG server using its API. This allows you to populate or update your knowledge base automatically, without manual intervention.

For this, we will use the SKKU ISS website as an example. The goal is to scrape the website for information and feed it into the lightRAG server so that our chatbot can answer questions about the SKKU ISS.

The code for this example is available in the pimatskku / iss-to-lightrag repository.

Create a folder called data in your project folder
Put the text files with urls into the data folder
Copy or download the code from scrape-and-index.py and put it into your project folder
Run the script

The indexing will take a while to complete. This is where we can take a break and let the code run in the background while we enjoy student’s presentations. Do not close the terminal, as it will break the indexing process.

At this point, your token usage may increase significantly. To monitor it, use the Metrics tab in your Azure OpenAI Studio.

Update your Open WebUI instance to the latest version

To ensure you have the latest features and bug fixes, update your Open WebUI instance to the latest version. You can do this by pulling the latest changes from the GitHub repository:

Depending on your setup, you might use one of the following commands:

pip install open-webui --upgrade

Or if you used the UV installation method:

uv pip install open-webui --upgrade

Then, start your Open WebUI instance:

dotenv run open-webui serve

After a successful update, you’ll be greeted by confetti and the “What’s New” page in Open WebUI, which highlights recent features and improvements.

Install lightRAG tool in Open WebUI

There are two ways to use lightRAG in Open WebUI: as a tool or as a model connection.

Implement lightRAG as a model connection

In your Open WebUI instance go to:

Admin Panel > Settings > Connections
Enable the Ollama API slider
Click on the + button behind the Manage Ollama API Connections label
In the URL field, enter http://localhost:9621 (the URL of your lightRAG server)
Press the rotating arrows icon to test the connection

Implement lightRAG as an Open WebUI tool

Open Open WebUI Tool - LightRag in a new tab

Then, in your Open WebUI instance, go to:

Workspace > Tools > Add Tool (+ icon)
Copy the code from the tool and paste it into the code editor in Open WebUI.
Give the tool a name, e.g., “SKKU ISS RAG” and a description, e.g., “Tool for SKKU ISS information retrieval using lightRAG”
Click on the Save and Confirm button to save the tool

Click on the gear icon to check on the available settings (valves) for the tool.

References

lightRAG:

Open WebUI:

Open Web-UI Release Notes

More to read:

SKKU

This post is licensed under CC BY 4.0 by the author.