Breaking it down yourself

Sometimes you read the plan and your gut says no. Not because any one step is wrong — but because there are fifteen of them, and you can already tell three are going to drift. The fix here isn’t asking for a better plan. It’s asking for a smaller one.

Never let a single run of work be longer than you can check. The longer the run, the more you have to unwind when something turns out wrong.

This is the one skill in this level that’s entirely yours. Plan mode and TODOs come from Claude. Slicing the work — deciding how big a bite to take — is your call, and it’s where a lot of the real skill lives.

Why smaller is safer

Think of each slice as a checkpoint. After every slice, you get to look, confirm it’s right, and then keep going. More slices means more checkpoints, and more checkpoints means that when something breaks, you’re throwing away one small piece instead of one giant one.

A big task done in one shot has exactly one checkpoint: the end. If it’s wrong, the whole thing is wrong, and you’re untangling all of it at once. The same task in four slices has four checkpoints — and a problem gets caught at the slice where it started.

Smaller chunks = more checkpoints = more reversibility. That’s the whole idea.

How to pick the first slice

Look for the first piece whose result you can see and check on its own. Often that’s the simplest piece. Very often it’s the reading piece — the part where Claude finds out what’s involved before changing anything.

Here’s the move with an example. Say you want to migrate a project from Tailwind v3 to v4 — a genuinely big, multi-file task.

• Bad first slice: “Do the whole migration.” Now you’re staring at changes across dozens of files with no idea if any of it is right.
• Good first slice: “List every file that uses a Tailwind class I’d need to change, and tell me roughly what would change in each.” Nothing got edited. But now you can see the surface area — and decide whether this is a tidy three-step job or a scary thirty-step one.

You learned the bones of this back when permission prompts taught you about blast radius: a smaller slice is a smaller blast radius per approval. Same instinct, applied to planning.

The thing that trips people up

Ambition. You want it done in one shot — one prompt, walk away, come back to a finished feature. It’s a nice fantasy. But one big shot that fails costs you far more than three small ones that each work. The slow-looking path — slice, check, slice, check — is almost always the faster one, because it doesn’t have a do-over hiding at the end.

What’s next

You’ve planned, you’ve sliced, the work is running. But work in motion can still veer off — wrong file, wrong approach, weirdly confident about something you know is false. The next lesson is what to do in that moment: how to interrupt, and why hitting the brakes early is one of the most valuable habits you’ll build.