abielzulio/chatgpt-raycast

ChatGPT raycast extension

chatgpt-raycast's Issues

Cannot select GPT-4

This is an awesome extension! This issue is probably user error, but after putting in my API key, I am only seeing two GPT-3.5 models to choose from. I generated a new API key just to test and I still couldn't get it to work. Is there another step I am missing? Thanks for all your work on this :)

Build not building to /dist?

Hey, super cool! Thanks for the work.

I'm new to raycast extensions, running npm run build which is supposed to... create a dist folder no?

The build command says built succesfully but no /dist folder ,and raycast says it can't find any?

Multiple Q&A support

Great work here! I'm wondering if there is going to be a multiple Q&A support in the future that uses the turbo API? Would be really useful to have that, and could be a replacement for the original ChatGPT client.

Request for Style Change

Hey Abiel !

Awesome work with the chatGPT extension with Raycast, I am currently using it now with a hotkey map and it is effortlessly amazing. Im making a request to remove or reduce the size of the bottom panel with the date, ID and conversation ID. It does not give the results enough space. And potentially allow the side panel width to be draggable to give the results even more space. Style the code blocks a little better with syntax colors and this extension will be 10/10.

Are you suggesting that I install it from the store? Because I can't seem to find it.

Noob question... How do I install it ?

Really sorry can't find the answer in google... how can i install it .. I thougt i did it I even enter my token and everything but when i try to use it the error " no js found" apear

Thanks

Feature Request: Add support for OpenAI Assistants

X-post of raycast/extensions#10082
Forgive me if this is a breach of protocol. The idea is to allow free text input into models (not just the pre-defined list in the dropdown) to allow configuration of the model names defined by the Assistant API.

Seems extension haven't used the API key and there's errors shown

I download the extension today and specify my account OpenAI key in the preference (which is already subscribed for ChatGPT plus)

But turn out:

• when I specify use of GPT-4 it shows GPT-4 is not supported
• If not specified it shows error 429
• my API key has shown the key is never used in my OpenAI account interface.

Guess the extension have not used the API key specified and would you mind looking into the problems and see if this is the case?
Thanks

429 error with Store version

Hi there,

I cannot seem to use the plugin, as I get a 429 error every time I submit a promt.

GPT-4 is not listed in the Model list

Hi, thank you for the excellent extensions!
I have enrolled ChatGPT Plus and have access to the GPT-4 model in ChatGPT web UI.
But when I try to switch the extension to GPT-4, it did not show up in the Models.

I have tried regenerate API key, reinstall the extension, restart Raycast. None of them fixed this.

Keyboard shortcut or alias for every model

So that I can use the clipboard and hotkeys to query for an answer using a certain model with just one action.

I custom API url, not work

I custom API url, not work：

It shows:

But, it works well on Chrome Extension (https://github.com/openai-translator/openai-translator):

Is there anything wrong?

Getting 429 Error

Getting 429 Error. From last 7 to 10 days.

When issuing new questions in the same conversation, only the response to #1 is remembered

Example:

Q: What's 1 + 1?
A: 2
Q: Add 1 to that
A: The result of adding 1 to 2 is 3.
Q: Add 1 again
A: If you add 1 to 2, the result would be 3.
Q: And add one again

A: If you add 1 to 2, the result would be 3.

Obviously the answers here should be 2, 3, 4, 5 -- but it seems to only remember the first question, 1 + 1, and every subsequent question only uses that for context, when it should be using the entire thread.

The ChatGPT website handles this correctly, continuing to add 1 to the previous response.

ChatGPT falied auth token

I've followed the instructions and I've updated the latest version of ChatGPT-api from transitive-bullshit and it still shows me the problem you see in the screenshoot. How do I do it now? Should I wait until you update this chatgpt-raycast? Thank you.

Models not showing on Ask Question command

I created some custom models, but I can't see them in Ask Question command. While in Conversations, I can use them. This seems to be a bug. Hope you could fix soon.

Some ideas about this extension

Hey,

Thanks for this great extension!

Here are some ideas you may consider:

• Allow user to custom the model temperature
• Automatic load the selected text as question (use getSelectedText API)
• #20
• Allow users to define custom prompts

Allow custom api url

for some well known reason, we need to set a reverse proxy to openai, so please provide a setting to override the api url

Use streaming mode to show the answer

Prompts support

It would be beneficial to have the ability to establish diverse prompts, similar to the functionality offered by chatbot-ui. Take the following feature as a simple example, that allows Chatbot GPT to perform code analysis.

Here’s my vision of the prompt feature:

• Users copy the source code from the editor or Github (during PR review, for instance),
• Users can then activate a prompt for the source code check using a keyboard shortcut,
• Users will paste the code into a dedicated text box,
• Prior to submission to Chatgpt, the system could then integrate the specified text into a preset prompt template.

The prompt of course could be anything.

Open chatgpt-raycast using hotkey

I assigned a hotkey to the chatgpt plugin in the Raycast settings.
However when I trigger it using the hotkey I get this error (presumably because it sends the question without input):

TypeError: Cannot read properties of undefined (reading 'length')
Main:ask.tsx:11:47

 8: }
 9: 
10: export default function Main(props: { arguments: Arguments }) {
11:   const [answer, setAnswer] = useState<string>("");
                                                  ^
12:   const [isLoading, setIsLoading] = useState<boolean>(true);
13: 
14:   const { question } = props.arguments;

Main:ask.tsx:11:47
zt:index.js:5:2490
    at ray-navigation-stack
Rr:index.js:5:2088

Is it possible to assign the plugin to a hotkey, while still allowing to input a question, without error?