Question: How to Provide `.gp.md` file as Context for GpChatNew

[harrisoncramer]

Hello,

Thank you for the great plugin! This is just a question.

I've used the GpContext command to create a .gp.md file in my repository that contains some information. However, when I run a GpChatNew command and ask the AI questions that are relevant to the context provided, it does not know how to answer. For instance, if I tell it my favorite color is red in the context file, the AI does not seem to retain that information in the chat.

This is using the default GPT-4 model.

Is there something I have to do to provide this context to every chat session in this repository automatically? Thank you!

[Mikkolehtimaki]

Hey did you figure this out? Running to the same question.

[harrisoncramer]

No I didn't

[nikbrunner]

I also just tried that unsuccessfully. I honestly was expected behavior for me: to automatically establish context when starting new chats.

Otherwise thank you for this Plugin! I really like it.

[Piotr1215]

Maybe someone will find it helpful. I'm using this workaround with a custom function on top of the .gp.md functionality per repository. It should be easy to create a wrapper around :GpChatNew and call it directly from nvim.

function gpt() {
    local current_path=$(pwd)
    nvim -c "GpChatNew" \
         -c "lcd $current_path" \
         -c "normal iKeep your responses short and simple, when asked to provide command, provide only one. Do not provide explanations unless explicitly asked for." \
         -c "normal o" \
         -c "call timer_start(100, {-> feedkeys('A', 'n')})"
}