Does CodeCompanion Set A Token Count Or Is This Provider Based?

[DennisTheMenace780]

Following up on my other discussion, i'm able to get some prompts going now. I have a pretty large prompt right now that seems to be exceeding a token limit.

[ERROR] 2025-02-07 12:59:07
Error: {"error":{"message":"prompt token count of 211289 exceeds the limit of 64000","param":"","code":"","type":""}}

It is unclear to me where this token count is being set, but I anticipate it is from the actual provider's API 🤔

Apparently Cursor does not have this limitation, but that might just be something that they're looser on than GitHub is.

[olimorris]

Yep this is a provider limit rather than a CodeCompanion one unfortunately. I believe Gemini 2.0 has a crazy limit and is super low cost compared to the others.

[DennisTheMenace780]

No problem, i'm still just learning about all the tech 😅 I'm working on playing with the documentation right now. Is this how you would expect to set the model given that copilot is the adapter? The model doesn't seem to be selecting claude 🤔

adapters = {
copilot = function()
	return require("codecompanion.adapters").extend("copilot", {
		schema = {
			model = {
				default = "claude-3.5-sonnet",
			},
		},
	})
end,
},