promptdojo_

Loops — read it, predict it, then trust it — step 1 of 10

promptdojo_›phase 01 · foundations›ch 04 · loops
lesson 1 of 3 · loops — read it, predict it, then trust itstep 1 / 10

Loops — what AI writes when you say for each

You will say a version of this sentence to Cursor at least once a day:

  • Go through every row in the CSV and...
  • For each user in the list, send...
  • Loop over the messages and find the one with...

Every one of those phrases triggers the same construct. AI does not write the same logic three times. It writes a for loop — a block of code that runs once per item in a collection.

Loops are the workhorse of every script you'll read this year. If you can read one cleanly, you can read 80% of the data-processing code Cursor will hand you.

The mental model: rebind, run, repeat

A for loop has three moving parts that fire in order, on every pass:

for pet in pets:
    print(pet)
  1. Rebind. The loop variable (pet) gets stuck on the next item in the source (pets). The first time around, pet is "luna". The second, "moose". The third, "biscuit".
  2. Run the body. Whatever's indented under the for line runs, using the current value of pet.
  3. Go around again. Until there are no more items, then the loop ends and the program moves on.

The thing to notice: the loop variable is just a regular variable. It gets rebound every iteration (remember the assignment chapter — "labels stuck on values"). After the loop ends, pet still exists, holding whatever the last item was. That's a quirk that bites people exactly once.

A worked example

The editor on the right has a 3-pet list. Trace what happens when you hit Run:

Passpet is...Output
1"luna"luna
2"moose"moose
3"biscuit"biscuit
—(loop ends)—

Same print(pet) line, three different outputs. Because pet points at a different value each time. The body of the loop is static text; the variable inside it is what changes.

Where AI specifically gets loops wrong

Two patterns to watch for in code Cursor writes:

  1. Using a too-generic loop variable. AI sometimes writes for x in users: when for user in users: would make the body readable. When you read AI code with single-letter loop variables over rich data, it's worth renaming them mentally as you go — x["email"] is harder to track than user["email"].

  2. Mutating the list while looping it. This is one of the most famous Python footguns. If you write:

    for user in users:
    if user["inactive"]:
        users.remove(user)

    …Python silently skips items. The loop's internal index moves forward while the list shrinks underneath it. Cursor still ships this in 2026 if you ask for "remove inactive users in the loop." The fix is to build a new filtered list with a comprehension, which we'll cover later in this chapter.

Run the editor. Three names, three lines, one loop body running three times.

read, then continue.