The quickest and simplest guide to spinning up a powerful local AI stack. Part 2 – SearXNG

In this part I’ll keep it short and sweet and just cover how to get Searxng going and connected to OpenWeb-UI.

So you want to comment out the following lines within the docker-compose.yml

cap_drop: - ALL

and then restart the stack:

python3 start-services.py --profile=gpu-nvidia

Once it’s all up, remove the comments and restart it again.

Now you can go to http://localhost:8080 and you should see the searxng homepage.

Now within OpenWeb-UI, go to Admin Panel then Settings, then web search.

Enter the following as the URL:

http://host.docker.internal:8080/search?q=<query>&format=json

Leave the rest as is. And you can test with or without the &format=json

Now hit the toggle to activate it.

Go to a new chat, and make sure you select a non-n8npipe model, such as llama3.2 and select the web search button.

Now just entry your question and success!

END TRANSMISSION