Work in TDD manner, or at least allow running tests

There are two parts for this request: the end goal and a simpler, mid-term feature.

The end solution:
In order to improve the quality of the generated application, I suggest that base44 will work in a TDD (Test driver development) fashion, like this:

• After getting the requirements from the user, the system will rephrase them as verifiable scenarios that can show the desired behavior.

• Then the system will generate automated tests (it can be anything from unit test to end-to-end tests. I can elaborate more on how to decide in private, as it’s a longer discussion) according to these verifiable scenarios

• The system will run these test and make sure that they fail only because the functionality hasn’t been created yet (if it fails on something else then the test should probably be fixed. If it passes, then it should consider whether the test doesn’t test what it needs to, or whether this functionality already works, and just notify the user that his request is redundant)

• Only then the system will generate the necessary code to make that test pass. It should also run all previous test to make sure that all of the previous requirements still work as expected (if not, try to fix, or notify the user about conflicting requirements…). The system can also refactor the code to improve efficiency or maintainability as needed.

I believe that this will strongly improve the quality and reliability of the generated applications.

As a mid-term solution, I could provide these guidelines as a system prompt (which I tried), but then I realized that there’s not way to run tests.

So the mid-term solution would be to allow writing and running tests.

PS: Test automation and TDD are my expertise (I’m also the author of the book Complete Guide to Test Automation). I’ll be very happy to help formalize this feature and answer deeper questions (e.g. like how to choose which kind of test to write for each feature…) as needed. You can reach my via WhatsApp, at 052-5327296.