# neat, so it doesn't intercept (or slow down) requests, it just alerts you if your agents leaked something? What's the token consumption like?
# oh I see in the readme it actually sits in the middle, so i guess the next question is what latency is like