[julienchastang]

I am not a meteorologist, but I've been using

http://arctic.som.ou.edu/tburg/products/realtime/eclipse/

and also

https://weather.cod.edu/forecast/

See "cloud cover products" from various models (e.g., GFS Monday 18Z).

Also see AFD discussions along path of eclipse at https://atlas.niu.edu/afd/

Unfortunately, with just a few days to go, the forecast for this eclipse in the US leaves much to be desired (much different situation compared to 2017). Your best bet appears to be northern New England and SE Canada, of all places. Some sections of the Midwest may also work centered somewhere around Carbondale, Illinois.