Initiating development of a chatbot with plumber and ngrok

August 3, 2017

(This article was first published on R, and kindly contributed to R-bloggers)

Chatbots have become a rage since some time now, with firms from various sectors investing in such bots that would reduce or remove the need of employing a call centre whilst maintaining similar levels of efficiency, if not greater. A notable example is DoNotPay, a ‘Lawyer Chatbot’ which is said to give free legal aid to refugees seeking asylum in the US, Canada, and some asylum support in the UK.

It may be that most of such companies are utilising in-house technology for chatbot development, it is also true that with the advent of developers like API.AI, Wit.AI, and with Facebook exposing their API and Microsoft providing their SDKs for Chatbot development, it has become much easier to craft an intelligent, engaging bot.

Although other programming languages like Python and Javascript have a fair amount of tutorials and packages that can directly or indirectly interface with Facebook’s API, those for R have been rather scarce.

Following are my limited, baby steps into the world of chatbot development using R and ngrok, making use of the Facebook Chatbot Development Platform.

Since we are using Facebook’s platform, we need to first set up a Facebook page, and then set up a Facebook app – both of these are easy steps, so I am going to skip covering them in detail. We create any page on Facebook, assign it a name, and we move on to Facebook for developers, where we create a new Facebook App ID.

Then, we add ‘Messenger’ as our product to the page, and generate a page access token that our chatbot will use later on.

To be able to send real-time notifications to our chatbot (and also to verify us!), Facebook uses Webhooks. We are required to list a URL here to which Facebook will direct any notifications we ask it to (keep your copy of Messenger Platform Document at hand, folks!).

But where do we host our app such that we can communicate it via its URL?
A possible candidate is Heroku – we can develop a simple chatbot and deploy it to Heroku. It will be live, and thus able to receive any verification requests and notifications that Facebook might send. R does not have an official buildpack for Heroku at the time of this writing, however. There is an unofficial buildpack on Github, but I have been unable to get it working for me and I am feeling lazy. So instead of heroku, I am going to rely on ngrok.

Set up Webhook and Complete our Verification

We need to first setup our Webhook, as Facebook will initially send a Verification request. It is only after this verification that Facebook will send notifications to our chatbot via the callback URL. In a verification request, Facebook sends 3 parameters:

  1. hub.mode (this is set to ‘subscribe)

  2. hub.challenge (this is a random string that Facebook expects us to send back)

  3. hub.verify_token (this is the same Verification Token that we will be entering in our Webhook’s setup)

Step 1: Write a script with plumber package

I am going to quickly script a few lines that will enable us to verify our callback URL, using the plumber package. Plumber allows you to create APIs by merely decorating your existing R code with special annotations, and I encourage you to read their documentation. I have, in my current project, found it to provide functionality similar to what flask does for Python.


#* @serializer html
#* @get /
myFunc <- function(hub.mode, hub.challenge, hub.verify_token){
if(hub.verify_token == "RRocker"){
 } else{
return(paste("Verification Token mismatch", 403))

Step 2: Save the Script and Run it!

Once you have copied the script shown above, save it as an R file, open an R session, and run the following code:

r <- plumb("")  


Step 3: Launch ngrok

Now, you need to open your ngrok and type in this code:

ngrok http 5000

This will show you a screen like the one below, where you can take note of the Web Interface and Forwarding URL. The web interface will allow you to inspect requests and their responses as they happen in real time, whereas the secure URL in the forwarding section will be used to register our callback URL with Facebook.

Step 4: Launch another R session and/or Ngrok Web Interface to test our script

You may open another R session to test our script. For example, I ran the below code and inspected the response in ngrok’s Web Interface as well as in R using the httr package:


url2 <- ""  

res1 <- GET(url2, verbose())

Keep in mind the significance of parameters shown above (after the question mark). You can see in a screenshot of ngrok’s web interface an inspection of our request and the response given by our script – things are looking good!

Step 5: Copy and paste the secure URL from ngrok to Facebook’s Callback URL in Webhook Setup

Here, we simply paste in the secure URL from ngrok to Facebook. If things are alright, we will see a green tick for the Webhook’s section.

What Now?

Admittedly, this is a very first step in our chatbot development endeavour. It is, however, a vital one. What we ought to do now is to
carry out a perusal of Facebook’s documentation to understand how we need to
handle POST requests, as it is such requests that will contain messages from other users as they try to engage with our chatbot.

I am unsure if I will be able to follow this through any time soon, as being employed full time and still working on a startup are already overwhelming tasks! But perhaps somebody more skilled will be able to take on from here sooner than I return!


This blog entry obtained a lot of help from Masnun’s blog post, which uses Python and ngrok for setting up an echo bot.

To leave a comment for the author, please follow the link and comment on their blog: R. offers daily e-mail updates about R news and tutorials on topics such as: Data science, Big Data, R jobs, visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...

If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.

Search R-bloggers


Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)