Find the gaps in your tests

Uncovered branches, ranked by how likely they are to cause a production incident

Below are my implementation and my existing tests. Find what the tests do not cover: 1) list all branches and boundaries, marking the ones with no case; 2) rank them by 'worst impact if it goes wrong in production' and give the top 5 cases to add (input, expected result, why it is easy to miss); 3) point out which existing cases are duplicates that could be deleted without losing confidence; 4) flag any test that asserts implementation details rather than behavior.

```
[paste the code and tests]
```

How to use

Add the top three first. Leave the rest until something actually breaks — do not try to write them all at once.

Good for

You are about to change core payment, coupon or permission logic and the tests were written six months ago.
Coverage says 78% but production incidents keep landing on boundary conditions.

Not for

The code is not settled yet (the interface changes three times a day): fix the contract first, or the tests chase it.
You only want the coverage number to go up: this prompt will tell you which cases are not worth writing.

Example

Input

def apply_coupon(order, coupon):
    if coupon.expired_at < now(): return order
    if order.amount < coupon.min_amount: return order
    if coupon.id in order.used_coupon_ids: return order
    if len(order.applied_coupons) >= 3: return order
    return order._replace(amount=order.amount - coupon.value)

Existing tests: one normal discount, one expired coupon.

What you get

List what is uncovered: amount exactly at the minimum, the same user's second use, stacking past the 3-coupon cap, a result of zero or negative, and the same coupon used by two orders concurrently. Ranked by production impact, the top five should add 'negative total' and 'concurrent reuse'; also point out that the expiry test asserts `_is_expired` was called, which tests implementation rather than behavior.

Common mistakes

Pasting the implementation without the existing tests: it cannot tell 'not covered' from 'not shown to me'.
Asking it to 'add all missing tests' and getting 40 low-value cases that cost more to maintain than they return.

Try tweaking it like this

Add 'give only the input, expected result and why it is easy to miss — no test code' to review the list before anyone writes it.
Add 'mark which cases need mocked time or concurrency' to surface gaps in your test infrastructure early.