Changing the approach

Did some "research". By that, I mean 1 prompt.

Essentially, there are 2 main problems with creating coverage for user flows.

Problem 1.

How do we automatically get all the user paths?

For linear flows, it seems simple enough.

Here is the trouble: dependent flows.

Owner flips toggle A, and it affects a part of a flow for a user.

Since the application is not aware of which toggle affects which part, it has to recursively re-run flows up to that point and check if a setting is exposing new branching flows.

Problem 2

The sheer number of permutations and combinations itself scales rapidly, and the main problem becomes prioritizing which ones to test.

What is the solution?

Instead of us asking the agent to do the mapping, we place a tracker on the website, it reads the actions and inputs, etc.

That way we get not only the actions and some what flows(I'll explain later) but we also get the frequency.

The frequency of the cases and each one's settings. Which solves the prioritization issue.

This helped me prioritize what I wanted.

My main problem currently is not breaking existing features.

However, there are still questions to be answered.

How do we define flows? When does a flow start?

How do we categorize flows?

For this, I will mostly use agents, but I need to understand first.

I already generated a tracker and setup api endpoints to create qa events, now I am setting up a Bubble.io test application to run it. Let's see.