normal trout
I'm thinking about building a VSCode Extension that would annotate strings in code (still thinking about how to filter the relevant ones) to show how many tokens will be used when the string is used as prompt and how much will it cost you to make the API call.
Would anybody also find this useful for estimating and working with prompts? How concerned are you about the prompt length when developing? Are you using prompts inlined in code or do you have some more elaborate systems for managing and testing prompts?
(Repost from api-support)