Slitherlink is a loop-drawing logic puzzle: you trace one closed line along the dots of a grid, and that single loop can never branch, cross itself, or touch its own path at a corner. Some cells carry a number from 0 to 3 showing exactly how many of that cell’s four edges the loop must use; blank cells can have any number of loop edges around them.
Every board on this page comes from a fresh random seed baked right into the URL, so a bookmarked or shared link always reopens the exact same grid. Pick a size above to begin: sizes run from a gentle 5×5 up through a demanding 10×10.
Density is what changes as the grid grows: a 5×5 board numbers 60% of its cells, so most clues interact with their neighbours immediately, while a 10×10 board numbers only half its cells, leaving longer stretches of blank cells whose edges you have to infer from how the loop must close around clues several cells away. The bigger the board, the more the loop’s shape has to be worked out from a chain of consequences rather than read straight off any one clue.

Slitherlink’s numbered clues are dug with the same uniqueness guarantee as a Sudoku’s digits: the generator only removes a cell’s number while an independent, from-scratch solver confirms no second loop could still satisfy every remaining clue. Every board on this page has exactly one loop that closes every 0–3 count correctly, reachable by deduction alone.
That proof is deliberately conservative: an inconclusive search is treated as if a second loop might still exist, so a clue is only ever removed once the search can rule that out directly rather than merely fail to find one. Clue density falls from 60% of cells numbered at 5×5 down to 50% at 10×10 (never sparser, since a sparser board hasn’t been proven safe within the search’s budget).