A two-string kite works one digit with two strings: a row where the digit has exactly two cells left, and a column where it also has exactly two. The strings are tied together in a box that holds one end of each — and from that knot, the two far ends fly out like the tails of a kite.
The two ends inside the box cannot both be the digit, because a box holds each digit once. So at least one of the strings must place the digit at its far end instead. Whichever it is, the cell at the crossing of the far ends’ row and column sees both of them — and can never hold the digit.
To spot one, collect conjugate pairs — houses where a digit is down to exactly two cells — and check whether a row pair and a column pair each poke one end into the same box. The eliminations are small but surgical: often exactly the cell you needed.
On the board
Follow the digit 4. Row 2 can only hold it at columns 1 and 4; column 2 only at rows 1 and 8. One end of each string sits in the top-left box — R2C1 and R1C2 — and that box cannot take two 4s, so at least one of the far ends, R2C4 or R8C2, must be a 4. The cell that sees them both is R8C4, at the crossing of row 8 and column 4: its 4 is struck, collapsing the cell from {4,5} to a bare 5.
This is the exact board the app's interactive Two-String Kite demo uses — step 33 of its 68-step walkthrough, drawn here with its live pencil marks.
