Hi Tomás,
In the final version of the book (reeeeal soon now :-) ) we'll have an appendix with a javascript example. We add unit tests as early prerequisites to the goal, driving a part of the graph.
When the test coverage is reasonable, you can go on making your changes as usual, with the feedback from the tests as covered in the rest of the book.
In addition to adding tests, you can of course make a manual analysis of the impact, do hyperaware editing and pairprogramming, manual testing etc to get even more feedback, but adding automated tests is recommended. :-)
Cheers
Daniel