How can I identify and verify a memory leak in a Gitter bot using its built‑in diagnostics?

I am running a Gitter bot that listens to room events and processes incoming messages through a custom handler. Over several days I notice the bot’s memory consumption steadily increases, even after periods of low activity, which suggests a possible memory leak. I want to confirm whether the growth is due to a leak and later verify that any fix I apply actually stops the growth.

I can only rely on Gitter’s supported capabilities such as the real‑time API, webhook payloads, and any logging the bot emits; I cannot attach external profilers or modify the host environment. I need a reproducible way to observe memory trends and to isolate the leak to a specific part of the code, then to check that a fix eliminates the unwanted growth.

What specific metrics or logs should I track to spot a leak? How can I tie rising memory to a particular handler or data structure? What validation steps confirm that a fix has resolved the leak without introducing regressions?