Coupons take off too much and tax is calculated wrongly

Done

/home/devdi/devswarm/evals/tasks/checkout_discount_tax/repo · 2026-11-09 16:09:14 UTC · took 37.7s

Plan (done)
Code (done)
Test (done)
Review (done)
Finish (done)

Cost

$0.0037

Tests

Passed

Iterations

Re-plans

LLM calls

Prompt read from cache

82%

Usage by agent

Cost per agent

Tokens per agent

Cache read
Cache write
Input
Output

View as table

Agent	Cost	Cache read	Cache write	Input	Output
planner	$0.0010	6,052	1,383	8	1,583
coder	$0.0009	9,562	2,600	8	990
tester	$0.0018	22,456	4,354	12	2,071

Timeline

planner iteration 1 Finished 3 calls · 9,026 tokens · $0.0010 · 10.2s

Loading…
coder iteration 1 Finished 3 calls · 13.2K tokens · $0.0009 · 7.6s

Loading…
tester iteration 1 Finished 5 calls · 28.9K tokens · $0.0018 · 18.8s

Loading…

verify iteration 1 Passed 0.5s

.......                                                                  [100%]
7 passed in 0.02s

Final diff

diff --git a/checkout.py b/checkout.py
index 2094fb6..b87dbc6 100644
--- a/checkout.py
+++ b/checkout.py
@@ -3,7 +3,5 @@ from pricing import apply_discount
 
 
 def order_total(prices: list[float], coupon_percent: float = 0) -> float:
-    undiscounted = sum(prices)
-    subtotal = sum(apply_discount(price, coupon_percent) for price in prices)
-    subtotal = apply_discount(subtotal, coupon_percent)
-    return round(subtotal + undiscounted * TAX_RATE, 2)
+    subtotal = apply_discount(sum(prices), coupon_percent)
+    return round(subtotal + subtotal * TAX_RATE, 2)
diff --git a/test_checkout.py b/test_checkout.py
index ea5239f..314afa2 100644
--- a/test_checkout.py
+++ b/test_checkout.py
@@ -6,5 +6,29 @@ def test_no_coupon():
     assert order_total([10, 20]) == 33.0
 
 
+def test_coupon_applied_once_and_tax_on_discounted_subtotal():
+    assert order_total([10, 20], 10) == 29.70
+
+
+def test_full_coupon():
+    assert order_total([10, 20], 100) == 0.0
+
+
+def test_discounted_subtotal_is_27_and_tax_is_2_70():
+    total = order_total([10, 20], 10)
+    assert round(apply_discount(sum([10, 20]), 10), 2) == 27.00
+    assert round(total - 27.00, 2) == 2.70
+
+
+def test_total_rounded_to_two_decimals():
+    total = order_total([9.99], 15)
+    assert total == round(total, 2)
+    assert total == 9.34
+
+
+def test_zero_coupon_matches_no_coupon():
+    assert order_total([10, 20], 0) == order_total([10, 20]) == 33.0
+
+
 def test_apply_discount():
     assert apply_discount(100, 25) == 75