[minor] This library doesnt seem compatible with VCR

[palladius]

Hi, I've tried to run VCR to intercept and cache Gemini calls, all good.
Tried with ollama and ollama-ai, the cache ends actually empty.
So I changed the code to a simple

  uri = URI('http://localhost:11434/api/generate')
  http = Net::HTTP.new(uri.host, uri.port)
  request = Net::HTTP::Post.new(uri, 'Content-Type' => 'application/json')
  request.body = { model: model_to_use, prompt: prompt, stream: true }.to_json
  ...

and it worked like a charm.

Newbie question, is there a way to intercept ollama-ai requests with VCR?
I'm pretty sure this can be done without changing your code, but a snipper would be helpful.

Riccardo