Auto-GPT WebUI 🎈

[choephix]

Hey guys!

I've been working on a super simplistic web application for the last couple of days, which wraps over Auto-GPT. I'm excited to hear what you think!

The spark was me wanting to reuse more than one group of settings for the AI to play around with it, without having to write them every time. I had a giant text file on my desktop and was just copy pasting yaml contents from there into ai_settings.yaml, so I thought, why not spend a few nights to make it a bit easier/nicer. This is basically the result.

I welcome any feedback and contributions my way. That said, this was mostly a one-off itch, I can't guarantee being able to maintain it much amid my other commitments, but I'll do what I can. ❤

Cheers!

---

Technical details:

The code itself basically comprises

• a React frontend with buttons to start, stop and interact with Auto-GPT as well as save and reuse multiple sets of ai_settings
• and a Node.js backend, which basically serves to execute the actual python script with the chosen arguments and to output the console log to the frontend.

---

🔗 Check out the repo here: https://github.com/choephix/auto-gpt-webui

[ResearchForumOnline]

That sounds really good thanks for sharing. Please update your install setup instructions so they are easier to understand:
Clone the repo, then enter the root directory and run the following commands:
put the command for that, or provide a better way, where is this going to be running from root or Auto-GPT folder or?

cd 'Auto-GPT'
su git clone https://github.com/choephix/auto-gpt-webui.git
npm install
npm run setup-auto-gpt

would that work?

[GoMightyAlgorythmGo]

A freeware installer that you can easily setup for users to use might be nice timesaver and increaser of engagement. Most people wont be as high motivation (biology different) i think wont do all thins required to try smth out not sure though. Having more and older people that are less tec savvy might be a asset. Also its nice

Also add snips/pictures might enhance the quickness of understanding and readme to be able to make a picture of it before installing, i think it could reach more people.

[ItsAhadun]

You can install it separately i think as it downloads Autogpt in its own folder separately so you dont need to install it in the Auto-GPT folder. The Instructions you wrote are correct

[JoneKone]

I like it =) thx

[kvleung]

i'm able to run auto-gpt command prompt version using Azure. But it seems that the webui doesn't read azure.yaml. Or is there anywhere I can set the deployment ID?

Traceback (most recent call last):
File "/Users/kvleung/Auto-GPT/auto-gpt-webui/auto-gpt/scripts/main.py", line 318, in
assistant_reply = chat.chat_with_ai(
File "/Users/kvleung/Auto-GPT/auto-gpt-webui/auto-gpt/scripts/chat.py", line 72, in chat_with_ai
relevant_memory = permanent_memory.get_relevant(str(full_message_history[-5:]), 10)
File "/Users/kvleung/Auto-GPT/auto-gpt-webui/auto-gpt/scripts/memory/local.py", line 102, in get_relevant
embedding = get_ada_embedding(text)
File "/Users/kvleung/Auto-GPT/auto-gpt-webui/auto-gpt/scripts/memory/base.py", line 9, in get_ada_embedding
return openai.Embedding.create(input=[text], model="text-embedding-ada-002")["data"][0]["embedding"]
File "/opt/homebrew/Caskroom/miniforge/base/lib/python3.10/site-packages/openai/api_resources/embedding.py", line 33, in create
response = super().create(*args, **kwargs)
File "/opt/homebrew/Caskroom/miniforge/base/lib/python3.10/site-packages/openai/api_resources/abstract/engine_api_resource.py", line 149, in create
) = cls.__prepare_create_request(
File "/opt/homebrew/Caskroom/miniforge/base/lib/python3.10/site-packages/openai/api_resources/abstract/engine_api_resource.py", line 83, in __prepare_create_request
raise error.InvalidRequestError(
openai.error.InvalidRequestError: Must provide an 'engine' or 'deployment_id' parameter to create a

[W3Warp]

Cool, where can I find some screenshots?

[turtlak69]

when i try to open the website, i can see "Connecting to server..." and this kind of errors repeating in dev console:

WebSocket connection to 'ws://localhost:2200/' failed:
connectWebSocket @ useWebSocketConnection.ts:16
(anonymous) @ useWebSocketConnection.ts:31

nodejs version v20.2.0
Python 3.10.6
Ubuntu 22.04

any idea?

[nrworld]

The UI lets me save the agent, but I didn't see any way to load it back into the UI

[ntindle]

Monitor page should have edit and such but we will be clarifying it soon