Thread Truncation Strategy

Hi, I have implemented my own version of threads and aiming for efficiency I summarize the whole conversation into a compact summary loosing all non important, redundant tokens. (just asking gpt to make a detailed summary) Then I resume the conversation using the summary and the newest interactions and summarize and replace the previous summary with and updated one again after the reply is generated this prevents the exponential growth in tokens to use for each run and keeps the number relatively flat from run to tun. I am defining the summary truncation strategy with a summarization prompt that tells the system what is important to keep in the context of the specific agent. Is that strategy something you could consider adding to the standard threads object?