Run data

Hook, skill or CLAUDE.md line? We ran all three (and nothing)

Claude Code 2.1.296 · 13 Oct 2026 · 12 runs · costs as Claude Code reports them. Raw summary (JSON)

Prompt

Add a new module src/lib/coupons.js exporting validateCode(code), which returns true for codes made of 4-12 uppercase letters or digits. Then add a discount(items, percent) function to src/lib/cart.js that returns the total including tax after taking the percentage off the subtotal. Add tests for both.

Runs

Run	Model	Cost	Turns	Time	tests pass	every .js file correctly stamped	Claude ran npm run format itself after its last edit	Diff
Nothing #1	sonnet-5-5	$0.113	15	29s	pass	pass	pass	diff
Nothing #2	sonnet-5-5	$0.111	15	28s	pass	pass	pass	diff
Nothing #3	sonnet-5-5	$0.106	12	25s	pass	pass	pass	diff
CLAUDE.md #1	sonnet-5-5	$0.082	10	22s	pass	pass	pass	diff
CLAUDE.md #2	sonnet-5-5	$0.082	10	18s	pass	pass	pass	diff
CLAUDE.md #3	sonnet-5-5	$0.086	11	20s	pass	pass	pass	diff
Skill #1	sonnet-5-5	$0.074	13	17s	pass	pass	pass	diff
Skill #2	sonnet-5-5	$0.089	12	21s	pass	pass	pass	diff
Skill #3	sonnet-5-5	$0.089	13	21s	pass	pass	pass	diff
Hook #1	sonnet-5-5	$0.098	14	23s	pass	pass	fail	diff
Hook #2	sonnet-5-5	$0.104	14	25s	pass	pass	fail	diff
Hook #3	sonnet-5-5	$0.100	14	26s	pass	pass	fail	diff