Hi folks, what would be the best
Hi folks, what would be the best practice to collect the complete number of tokens used in an agent run that uses multiple tools (and each tool has ability to use more tools as well)? I am running heuristic tests with Jest, but the response from the primary agent call only seems to return the number of tokens used by ai.Generates in that first layer. I was thinking about running the tests against a running dev server to get the traces, but there's no info about the usage there as well.
