In order to apply a filter the way that you're describing, you'd have to have the method name available as a log context property that you can filter on. There's nothing built-in in Serilog that captures method names, thus you'd have ensure this property gets created when you're writing to the logs.
Using LogContext you can add a property that you know about, which would flag the log messages to be included (or excluded) by a particular filter.
e.g.
using (LogContext.PushProperty("ThisIsSuperCritical", true))
{
Log.Information("This message will have the property `ThisIsSuperCritical`...");
Log.Warning("... and this warning too");
}
Then in your logging pipeline, you apply a filter on the sub-logger based on the property you're looking for. E.g.
Log.Logger = new LoggerConfiguration()
.MinimumLevel.Verbose()
.WriteTo.Logger(wt => wt
.Enrich.FromLogContext()
.Filter.ByIncludingOnly(Matching.WithProperty("ThisIsSuperCritical"))
.WriteTo.SomeSink())
.CreateLogger();
N.B.: Don't forget to add the Enrich.FromLogContext() otherwise the properties won't be added to the log context.
For a small number of methods the above is fine. If you have many methods and need to get the name of the method dynamically, then you'd have to use a custom enricher that reflects over the call stack to capture the method name (which might have an impact on performance as it's an expensive operation).
