Best approach to reduce prompt length

I am trying to generate 100s of articles using different combinations of a dozen or so keywords list I have. Each keyword list is more than 50 comma separated phrases. I must send these keywords lists otherwise I don’t get the desired output. My current prompt is going over 1000 words because of these lists and costing me a lot. I was hoping there was a way I could “store” these keywords lists with the model and then just provide keyword list name so it would automatically pick from those keywords.

For example I need articles on the topic - <sports> <country> <age_group>, so basically I have keyword lists for sports, country names and age groups and I am able to generate articles such as How Teenagers play soccer in the USA, How teenagers play soccer in UK and so on…

I have seen this post that discusses a similar problem, but I am not sure if fine-tuning is the right approach because (a) I don’t have sample articles on each and every topic. (b) even if I have the samples I fear missing certain keywords or entire list.

So what’s the best way to reduce my prompt by storing my keywords lists somewhere in the openai models. Thanks.