Slitherlink

Overview

Slitherlink presents numbers inside a grid and asks you to draw exactly one closed loop around them. A number states how many edges of its cell belong to the loop. A blank cell imposes no count of its own. The line cannot branch, cross itself or split into multiple smaller loops, so local edge decisions must cooperate with the final global shape. The hosted Loopy engine includes different grid types as well as sizes, although a small square grid is the clearest starting point. You are drawing along edges, not shading the numbered cells. A nearly complete board can still be wrong if two tidy rings have formed independently or if one dangling line end has nowhere left to go. Exclusion marks carry half the logic, so a good solve records what is impossible as carefully as what is certain.

How to Play

Click an edge to include it in the loop. A secondary click marks an edge that must stay empty. Enable Secondary on a phone when placing those exclusion marks, then switch back to draw lines. Work around the numbered cells until the required edge counts are met and the path closes into one continuous loop. Use Undo to correct an inaccurate edge tap. The Type menu changes the board configuration; Restart removes your moves on the current puzzle, and New generates another puzzle. Marking excluded edges is never wasted effort, because a cell with two exclusions and a count of two has its remaining edges decided.

Tips & Strategy

A zero excludes every surrounding edge, which often forces neighbouring threes. At every vertex, the loop must use either zero or two incident edges; a single line arriving at a vertex may force its exit. Track these vertex constraints alongside the cell numbers. Avoid closing any small ring before the rest of the required edges have joined it. On irregular grids, count the actual sides of each cell rather than applying square-grid patterns from memory. When you get stuck, look for an excluded edge that leaves a vertex with only one possible continuation.

Classic Clue Patterns

Several configurations are worth learning by name. A 0 is the friendliest clue in the game: exclude all four surrounding edges immediately, then check which neighbours just lost options. A 3 in a corner forces both outer edges of the corner cell. The reasoning is a small proof: if one outer edge were excluded, the other would have to be present to reach the count of three, and that edge would dead-end at the corner dot with no way out. Next, a 0 beside a 3 removes their shared edge, so the 3 must use its other three edges in full. After drawing those, the vertex rule takes over: at each end of the new path, exactly one continuation remains possible once other exits are excluded, and a corner 3 often unspools into a long forced run along the border.

Worked example: a corner 3 shares its inner edge with a 0. Draw the two outer edges, exclude the shared edge, and the 3's remaining inner edge joins the set. The path now reaches the two dots beside the corner, and because a line may not end, each must continue — if one of those dots touches another excluded edge, its exit is forced too. Chains like this are how a single corner opens an entire border.

Habits to Unlearn

The most common failure is closing a neat little ring in one region while the rest of the board still needs it; the single-loop rule punishes early closure. Guessing without a fallback is the second trap — place an exclusion mark for the alternative before testing a speculative edge, so Undo stays clean. On triangular or hexagonal grids, resist importing square-grid intuition and count each cell's true sides. Finally, remember that blank cells still shape the loop through the vertex rule even though they show no number.