Casper
← Leaderboard

weekly-brief

A short Monday-morning brief comparing last week's sales to the week before, with the products that moved.

Latest eval run

Passed 100%Cost / run $0.022Time / run 6.7sclaude-opus-5-5Oct 17, 2026
  • PASSlast week on demo data$0.026 · 6.9s · 3 tool calls · 0 proposals
    Net sales for the week of Oct 5–11 were $32,479, down 1.1% from $32,834 the week before. These are demo figures, since no real store is connected.
There were 209 orders, down 13 (5.9%) from 222. Average order value was $161.56, up $7.50 (4.9%) from $154.06.
Packable Rain Shell: 71 units, $9,159
Everyday Daypack 20L: 91 units, $8,099
Ultralight Hammock: 79 units, $5,846
The biggest change was fewer orders, but each order was worth more, so net sales stayed close to the week before.
  • PASStwo weeks back, five products$0.019 · 6.5s · 3 tool calls · 0 proposals
    Weekly brief for Sep 28 to Oct 4, 2026, compared with Sep 21 to 27. This uses demo data, not a real store.

Net sales were $32,834, down 3.1% from $33,897 the week before.
There were 222 orders, up 5.7% from 210, but the average order value fell 10.0% to $154.06 from $171.27.
Packable Rain Shell: 78 units, $10,062
Everyday Daypack 20L: 77 units, $6,853
Ultralight Hammock: 85 units, $6,290
Trail Runner Tee: 82 units, $3,116
Insulated Bottle 24oz: 77 units, $2,618
Net sales fell even with more orders because each order was smaller, though refunds dropped from $2,069 to $1,367.

The prompt

Weekly brief

You write the brief a store owner reads on Monday morning in under a minute.

Steps

  1. Work out the Monday-to-Sunday range of the week weeks_back weeks before the current date (the current time is given to you in the store's timezone), and the same for the week before it.
  2. Call get_sales_summary once for each of the two weeks.
  3. Call get_top_products for the reported week with limit set to the top input.

Output

Plain text, no markdown. Five to eight short lines:

  • Line 1: net sales for the week with the percentage change against the week before.
  • Line 2: order count and average order value, with the change.
  • Then one line per top product: name, units, revenue.
  • Last line: one sentence on the single most notable thing in the numbers. If a tool result says "truncated", say the totals are partial.

Every figure comes from a tool result in this run. You may compute differences and percentage changes between two tool results, nothing else.