timmaugh said: I'm going back and reading what you've written, and it sounds like you're writing your own script. If so, you have a couple of hurdles: 1) listen to all chat messages, but differentiate which ones contain the rolls you care about 2) take action only on the messages that matter, and only on the rolls that you need within that message 3) determine to which attribute you wish to apply any given roll (if the message contains more than one) 4) get the roll value 5) apply it to the attribute To answer your specific question of getting the value from an inline roll (in a javascript context, anyway -- hopefully that's what you need)... you'll want one of 2 approaches. If all you need is to extract the value, you can use this snippet: msg.content = _.chain(msg.inlinerolls) .reduce(function(m,v,k){ m['$[['+k+']]']=v.results.total || 0; return m; },{}) .reduce(function(m,v,k){ return m.replace(k,v); },msg.content) .value(); I broke this down and explained it in this post , if you wanted a little more information about how it works... but basically it looks for any roll marker ($[[0]]) in the command line and replaces it with the value of the roll. This changes the command line. Again, if you are listening to all messages (including non-API), then the message you are working on will have already hit the chat output. However, if it is an API message, then you might want to take measures -- like not acting on a message intended for another script, or not rewriting the command line of the message object but just writing the new line into a variable of your own. The other option you can use is to hook into libInline , which will not only give you the value of a roll, but also the breakdown of dice, or tables, or even the HTML of the roll tip for a chat output. It just depends on how much you need to do with the rolls once you have them. I've tried this, it works but it bugs out sometimes and picks the wrong roll. I've decided to cut to the chase and have my own dice rolling directly inside the script, it takes number of dices, number of faces and some math.random(s) later I have my result. I know math.random isn't really random, but I've asked a professor in uni (I'm studying computer science) and he says it's good enough since I also randomize the use of ceil and floor to approximate the random number, thanks for all the effort though, not only from timmaugh but from everyone else! Loving this Roll20 community