I worked on a library to do this for my own analytics thing[1]; adding a CLI so you can filter bots from logfiles isn't too hard: just need to parse the log lines to a http.Request{} with the correct fields filled in (mostly User-Agent, but also looks at IP address).

One of the signals is from various JS properties though (navigator.webdriver, window._phantom), so you'll miss out on that. I have some other ideas as well.

Either way, it's better than nothing?

https://github.com/zgoat/isbot

You could do this with nginx for sure. I don't know what the impact on performance would be.

Or even a logging module in the web server itself...