> ## Content Index
> Fetch the complete content index at: https://www.waterboy.tech/llms.txt
> Use this file to discover other available public pages before exploring further.

# Coverage based on Data fields
- URL: https://www.waterboy.tech/coverage-based-on-data-fields/
- Published: 2026-10-20T16:49:37.000Z
- Updated: 2026-10-20T16:49:37.000Z
- Author: waterboy
- Tags: QA Monk

The progress on the next versions for the test case generator has not been great. 

Seems like I'm endlessly updating a prompt without structure. 

Then came across a different way to approach the test case generation and process mining. 

## Old solution

Currently, workflows are mined, and all branches are created. 

I still don't know how this works, and it seems like a black box. 

[testcaseGenv2.2Initial old solution filetestcaseGenv2.2.md11 KBdownload-circle](https://www.waterboy.tech/content/files/2026/10/testcaseGenv2.2.md "Download")

## New solution.

I figured the main cases always touch database fields.

We can explore the usage of each field in workflows and then map the flows. 

This gives us progress markers and a touchpoint to generate cases. 

Say a new feature touches 5 fields. 

All past test cases that use these fields need to be retested. This quantifies regression.

It also eliminates all the fragile frontend cases that have infinite possibilities. 

Like the user fell from a tree and opened both popups.

Which doesn't seem like a good thing at first. However, frontend cases seem to be generated at random(non-deterministic).

And most front-end cases are not that important. 

Now at least I have a smaller problem to focus on.

This allows the generator to create deeper combinations and permutations.

So here is the version of the test case generator.

[testcaseGen2.4testcaseGen2.4.md10 KBdownload-circle](https://www.waterboy.tech/content/files/2026/10/testcaseGen2.4.md "Download")

Already, my cases have been reduced. 

\[ \] Now I just need to verify the Quality.

I also made multiple updates to the HTML display page. 

![](https://storage.ghost.io/c/19/ff/19ff0f2b-7d87-4f38-8a35-910b9e637720/content/images/2026/10/Screenshot-2026-10-20-at-10.09.48---PM.png)

A section for marking individual roles and variants that cover the same test case.

![](https://storage.ghost.io/c/19/ff/19ff0f2b-7d87-4f38-8a35-910b9e637720/content/images/2026/10/Screenshot-2026-10-20-at-10.10.44---PM.png)

Above is the raw list of fields that need to be validated after a test case is completed.

For now, it can be done manually, but later I will use the agent to auto-check these fields.

The core hypothesis is:

At the end of the day, if the data in every field is correct, 80% of our problems should be solved.