Jamie Taylor

What if this happens?

What if that happens?

Which is weird.

Yeah, absolutely.

Absolutely.

And then I'll do the spec, then the plan, then implement.

But the key thing, and this is a pro tip for anyone who wants to try out SpecKit, is if you're doing it that spec, so you do spec, plan, implement.

clear the context between each one, right?

Because right, especially in the middle, when you come to do the plan, it's going to do a whole bunch of research.

It's going to write a whole bunch of markdown files to do with like what it thinks the data model should be, what it thinks the contract should be, which libraries to use, where to go to get the library information for the API calls and stuff.

So it's going to create you almost like a wiki just for this one feature.

So you want to clear the context every time you complete one of the steps.

And then the other thing you can do is you get to the break into tasks step, which is just called tasks.

It then ships with a command to analyze everything it's done so far and look for yet another set of gaps in user stories or gaps in knowledge.

So you can have it check its homework, if that makes sense.

So it does both.

So what it does is it goes back to the original spec, reads through it to double check that there aren't any more gaps in those user stories or assumptions.

And then it goes, cool.

So based on this spec, have a look at the plan.

Does the plan match the spec?