Using Puppeteer to Train UP9

UP9 has several modes of operations. It can be deployed 'close' to each services (e.g. as a sidecar in K8s), however, it can also be trained to observe workflows and automatically create tests that cover those workflows.

With this option, installing UP9 on the target system isn't required. Puppeteer opens a browser window and provides access to the traffic that is generated from simply using the target system through the Puppetter browser.

Once you have the UP9 CLI installed, you are ready to start observing workflows using Google's Puppeteer.

Sign Up & Install

To install, please sign up using this link: You can ignore the on-screen instructions for now. To use Docker-Compose, you need to install the UP9 CLI.

Install the UP9 CLI using NPM:

Copy to clipboard
npm i -g up9

Or, with Brew:

Copy to clipboard
brew tap up9inc/brew && brew install up9

Assuming your browser is open and you are logged in to your UP9 account, authenticate the UP9 CLI:

Copy to clipboard
up9 auth:login

The CLI will use the browser to complete the authentication.

Launch Puppeteer

The following command will open up a browser window and record session traffic into a new or an existing workspace. From command prompt, run

Copy to clipboard
up9 tap:start

We suggest you use a meaningful name instead of ‘,’ as this will be the name of the new workspace. If you'd like to add workflows to an existing workspace, simply use the name of an existing workspace.

This command will open a new Chromium browser window. Use this browser window to go through the workflow you'd like to observe. All steps made in this browser will be recorded and sent for processing. Puppeteer Browser Window

Allow permission to Chromium when prompted. As you browse, you will see a requests counter (in separate tab) incrementing, which indicates application traffic is being recorded and sent for processing.

When done, close the Chromium browser and go to your UP9 dashboard to view your new (or existing) workspace.

Model Accuracy

UP9 uses machine learning to create a model based on the recorded traffic. The more passes found, the stronger the confidence level of the model. In layman's terms, it is best to record the same workflow several times into the same workspace to make the model more accurate.