I don't know if this is how things will work in future, but if anonymous users, like IP users, would have Flow enabled by default, Huggle would no longer be able to retrieve / parse list of warning templates that were already sent to user in order to know what warning level they have.
To make this easier to understand for people who don't know much about vandalism or Huggle: the classic schema for wikis like enwiki is that vandal user is supposed to receive 4 warnings before they get blocked and each time the warning level increases, and the template used to warn target user is more direct and less friendly. Now, for this to work, Huggle needs to know what current warning level of user is, before it attempts to warn / block them. Right now, what we do, is that we retrieve the content of talk page and parse it using special "magic strings" that are defined in project settings and using these we determine highest level, recent warning template.
If user had flow enabled on their talk page, this parser probably wouldn't work as there would be no way to retrieve the talk page in standard wiki-code format. It's probably necessary to again override functions that retrieve the talk pages, check if user has flow enabled, and if yes, then analyze the contents of page using flow APIs.
Because vast majority of all vandals are either IP users or very new users, probably close to 0 of them in this moment have flow on their talk pages, so this task isn't really high priority, but this may change in future if it was decided to enable flow by default for all new or IP users.