Reports become test cases the matcher is graded against before an update ships.
Delvefall isn't a chatbot guessing, it compares the printed rules text of every card with plain math,
and examples like yours decide which differences should count.
Witching Well 100%
$0.18#11085
"{3}{U}, Sacrifice this artifact: Draw two cards."
Sewer-veillance Cam 100%
$0.20#11092
"{3}{U}, Sacrifice this artifact: Draw two cards."
Mnemonic Sphere 97%
$0.25#11461
"{1}{U}, Sacrifice this artifact: Draw two cards."
Courier's Capsule 93%
$0.26#20182
"{1}{U}, {T}, Sacrifice this artifact: Draw two cards."