The goal of quickchat is to give me quick access to an LLM while I'm in Positron while I wait for someone more competent than me to implement it directly.
This package will thus not be maintained long term as I fully expect it to superseeded.
You can install the development version of quickchat like so:
pak::pak("emilhvitfeldt/quickchat")
there is only one function quickchat()
and it brings up a shiny app that uses shinychat and ellmer to let you interact with an LLM.
3 llms are supported:
ellmer::chat_claude()
ellmer::chat_openai()
ellmer::chat_gemini()
And they will be available based on whether their envirment variable is set or not.
quickchat::quickchat()
{
"key": "cmd+. cmd+5",
"command": "workbench.action.executeCode.silently",
"args": {
"langId": "r",
"code": "quickchat::quickchat()",
"focus": true
}
}