With https://github.com/alacritty/termbenchbot there are now reliable daily benchmarks of the latest Alacritty master, however this does not help with benchmarking specific branches and pull requests.

Ideally, it would be possible to run benchmarks on demand, so that Pull Requests can be easily tested without requiring any manual performance validation.

One could of course just build a simple GitHub bot and then start the benchmarks, however running the bot and the benchmarks on the same machine requires the benchmark machine to be online 24/7, which it currently is not to save some energy. So to make this possible, one would either have to find some way to request benchmarks externally using a low power raspberry pi for example, or commit to running the benchmark machine continuously (which might come with a significant cost).