samuraigpt/open-custom-gpt

Create Custom GPT and add/embed on your site using Assistants api

open-custom-gpt's Issues

Possible to share with users without them needing to add open ai key?

Hello,

I love the idea of this tool, is there a way of sharing the Assistants with my users without them needing to put their own openai key in?

suggestion - sample to programmatically update knowledge base

I customized my openapi gpt with some data - then hit a wall when I wanted to programatically update the knowledge base. perhaps the sample would be fit a lot of users - use-case for using this repo too.

How can this be done with this repo?
I guess there's many options - do we have to use pinecone? or is there a way to do a custom scheduled tasks to replace a file?

I guess thinking out loud - I kind of want a page - admin panel that just helps with this kind of steps - or spoon feeds user on this use case.

Not showing any answer

Is there a way to see existing Assistants in ChatGPT?

I was hoping that by adding my API key I would be able to see list of existing Assistants, do you know if this is possible?

Deploy in Vercel?

Great app! Thank you for sharing this in the community.

Is there instruction on how to deploy this using Vercel?

Replit code

is it possible to have the code that can run in replit? Then I can also create the readme for the replit version, but a code that is working properly is essential obv

Adding an upload button

I miss an upload button in the chat UI. I would like to see this button and the function to upload files and that the AI can analyse those files and you can chat about it to get advice or feedback for instance. I'm aware this will cost you more tokens, but due to extra functionality you can calculate this in and take that into account when you would like to monetize it. 😁

Issues running npm run start or npm start

I get an error while running the application in a production environment on my server:
(previously, I have run on the server: npm run build)
npm run start or npm start

I am attaching some screenshots.

Copy Embed doesn't add the protocol (http(s)://)

After clicking the copy embed and pasting it, local URL came through in the iframe as localhost:3000/embed/asst_H25Y854Nuqovy5VqlFFetnPD, I had to add the https:// to get it working.

Voice Interface available

Will it be possible to create an assistant that has the voice interface enabled similar to what I can do in the mobile iOS app? I have created a custom GPT that is an AI Sales Trainer and I want to recreate it outside of the chatgpt website using the assistants API.

Can add proxy domain functionality？

Hosted version of Open Custom GPT not working

The 'Hosted version of Open Custom GPT' is not working.
When entering all the information (name, instructions, file,...) and clicking on "submit" nothing happens.
There is also no field to enter the OpenAI API.

Is ai-plugin.json not used and description (prompt) is not take effect for custom plugin?

I run my plugin locally and use in playground, but it seem that there is no api call to ai-plugin.json. And the description in both yaml and json does not take effect. Do I miss anything?

UI Enhancement

Hey the UI of the online hosted web page is not up to the mark, I can make the following changes for now:
-Centre the content for better user experience
-Creating a Home Button
-Perfecting the background
-Providing colour to the buttons

Relation to customgpt.ai ?

Sorry for the confusion, is this project similar to customgpt.ai. Or related in any way?
Is this completely free? I red something about 90usd per year, what is that for?

Way to expensive

I just made a custom GPT and looked for options to embed it into my website and found this. At first, I was happy because this was one of the only functioning things I found. Then, I looked at the pricing and found out you have to pay 90 dollars! That is way too much for something like this. If you ask me, this programmer should lower the price for something that is very slow.

Replit code

API key request using other devices

Is there a way to set the API key as fixed so that external users can interact with the chat?

I've installed it and it works perfectly in Chrome; however, when I open it in another browser or device, it asks me to enter the API KEY.?

Assistants isn't loading

Hi,

downloaded a HEAD version, then npm install, insert api key, but on next screen have just create assistant so it's not list already created ones and if I click on "Create new one" I get

Application error: a client-side exception has occurred (see the browser console for more information).

In console:

Uncaught SyntaxError: redeclaration of var Audio 749-437c88a29d01f3f9.js:9:22911note: Previously declared at line 6, column 23254749-437c88a29d01f3f9.js:6:23254

ChunkLoadError: Loading chunk 749 failed.
(missing: http://127.0.0.1:3000/_next/static/chunks/749-437c88a29d01f3f9.js)
NextJS 40
j
e
e
n
t
t
promise callbackU
createFromFetch
promise callback7950/t.createFromFetch
f
E
rb
iU
o2
oB
t
oB
C
1756
s
8261
s
4417
s
4887
s
4040
s
811
s
2019
r
r
2019
s
n

O
c

472-69984e29036e4540.js:1:3763

Have some clue, what I'am doing wrong?

Thanks

API Keys Exposed Publicly in Shared Links

Hi, the current sharing mechanism reveals API keys publicly in shared links, posing a critical security threat. I think we need a new, secure linking method.

I have my own custom gpt's in open ai , and now I want to use it through langchain .I don't want to use gpt 3.5 or gpt 4. I want to use mt custom gpt

One question

If i create an assistants using this , can i see it into my ASSISTANT DASHBOARD ?

Internal Server Error when trying to signup on customgpt.thesamur.ai with google

Up/ down option , When a contract gets knocked out it won't take that contract out leaving the customer in a position they can't purchase the cheaper option because it's not on the list of contracts

_

Build - OpenAI package issue - Uncaught SyntaxError: Identifier 'Audio' has already been declared

Hi,

Great projects, thanks! Was trying to do a production build of the next app, and used output:export option to provide just static files to serve via S3 (my use case allowed that, the assistant was already created and tweaked the app a little bit just to use embedded route). Anyways, was struggling after building with Uncaught SyntaxError: Identifier 'Audio' has already been declared error in the console. Found this gem:

openai/openai-node#603

FYI solved it by bumping the next package from 14.0.3 to 14.0.4 so we might do it also in the repo :)

I pay and not work

I pay with stripe, and the app not work. I need cancel my subscription now.

Cannot login customgpt.thesamur.ai

Login with google, then redirect to localhost:5000.
Seems to not configure google login correctly.

Is it possible to save multiple agents?

I seem to only be able to generate one agent at a time. Will there be added functionality to manage multiple agents?

spanish language

I want to know how to configure it so that it speaks to me in Spanish. Ask in Spanish and answer in Spanish. I don't understand why if I ask in Spanish he answers me in English

Runtime error when trying to Submit with Actions copy-pasted to Functions

Hello, I fill the Function frame by copying and pasting a JSON API schema that works fine in my original Custom GPT, I mean from its Actions. I'm getting this error every time, whatever Function/Action I paste, so it is not possible to Submit with that code... Maybe a wrong understanding of Functions option from my side? Please kindly advise, thank you!

The app keeps asking for API Key every time the page is refreshed

Hi. Thank you very much for this projects, it's very useful.

Everything works fine, but there seems to be annoying bug - every time I refresh the browser I see the form asking me to provide OpenAI key, even if I provided it before and I already have a working assistant chat.

Any ideas how to prevent it?

Help with hosted variant

Signed up for 7 day free trial - no option to cancel anywhere on site, or any email?

Formatting Text Response

Feature Request

• Add various text formatting conditions or remove weird formatting returned by ChatGPT

Issue

ChatGPT will occasionally return text encased within ** to denote bolding. Would it be possible to either filter this out from the response or bold text accordingly?

Other weird formatting issues resulting from italics and other formatting. ChatGPT will return text surrounded by * on either side if it's italicized.

Additionally, some text would look better if it was allowed to appear with line and paragraph breaks.