Setting the endpoint parameter breaks the extension.

[lukolszewski]

If the endpoint parameter is set to anything including OpenAI's current endpoint" https://api.openai.com/v1/chat/completions" The extension fails to work. It doesn't even attempt to talk to the address provided, just errors with:

Error: Error accessing OpenAI. The "Update OpenAPI Api Key" command may help you.
undefined
	at [/home/luk/.vscode/extensions/jaaxxx.llm-book-0.1.1/dist/extension.js:1:4704](https://file+.vscode-resource.vscode-cdn.net/home/luk/.vscode/extensions/jaaxxx.llm-book-0.1.1/dist/extension.js:1:4704)
	at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
	at async Object.<anonymous> (/home/luk/.vscode/extensions/jaaxxx.llm-book-0.1.1/dist/extension.js:1:2494)
	at async n.$executeCells (/usr/share/code/resources/app/out/vs/workbench/api/node/extensionHostProcess.js:143:96089)

I know this is not an API key issue as when I delete the endpoint it works fine. This also prevents use of any local OpenAI api endpoints.

I don't know why would it fail as the code seems to set the endpoint to the same value, but it fails nevertheless.Screenshot attached.

For some reason chaning the parameter in the settings GUI works fine, but using the Configure LLM button and inputting wia the command box results with the error above.

Furthermore, if it was possible to set this (as well as other settings and parameters) on a per document basis that would be great.

[JacksonKearl]

The Configure LLM menu is for adjusting the fields included in the body of the request. I see the PR you raised.. including a settings field in the top-level .llm.json format is a possibility.

[JacksonKearl]

Pushed a new version which fixes the error message not being properly preserved.