Create /ask command

[Wikijito7]

Sends the prompt to AI's API and generate a tts sound with the answer given.

As of right now, this will be using HuggingFace API, as OpenAI have closed their free credits program. Now, this ask command is working with a chat interface, so we could add some extra functionality like addind a flag to convert current thread into a chat with whichever model we are using.

Given prompt should be displayed inside the thread.

Mantain support on the context menu interation (right click a message) in order to generate a response with the content of given message.

We have to check how to stop using huggingface library and transform the request into an api call.

• Create ask command and it's logic
• Use ktor to request and get the response
• Pass along the response into tts class
• Create context menu interation

[Wikijito7]

Check this post https://blek.codes/blog/duckduckgo-ai-chat/

[Wikijito7]

Check these two things:

• https://huggingface.co/docs/api-inference/tasks/chat-completion?code=curl
• https://huggingface.co/docs/api-inference/tasks/text-generation

[Wikijito7]

Also, new endpoint with llama3 (needs to be worked with a pre-prompt)

https://ai.ince.pt/projects/demo1/chat

{"question":"blablabla"}

Needs a basic token auth. We can get it from the demo webpage.