ToL flags English text as good or bad. It looks for ease of reading, thoughtfulness, and truth.
I use ToL to read news, to read what I write, and to read email.
Keep an open mind when interpreting the results. Don't be alarmed if text written by good people sometimes gets flagged as bad. This isn't a bug. Most times the reason this happens is that the author was trying to sound impressive.
I don't know yet if this is bad but it's hard to pay attention to text now once it's been marked as red. My eyes skip the red and land on green. If I stop on red it's out of curiosity to see if the AI works well. I hope this doesn't come at the cost of not reading text that didn't get highlighted at all even though it's valuable.
The accuracy needs work. ToL can use a larger training set than the modest set I used to seed it. Some phrases also have meaning that ToL can't understand yet. The tough part in both cases isn't just to ensure no loss in quality. It's to choose what I want the AI to do.
The biggest problem with the browser plugin is that it's not easy to change text in HTML. Many sites don't use HTML tags properly. The plugin can't simply change the background color in HTML. It has to remove and reinsert HTML, and sometimes it loses the formatting along the way. As a result the plugin works only on
I stopped using ToL after a few months when Firefox wouldn't let me load it through "about:debugging". I don't know exactly what happened but I think it was because of a bad software update. A year later it fixed itself. And two weeks after that it broke once more after a software update. 
ToL processed 668,229 API calls since August 2016.