Mine Sweeper is a pure logic puzzle — no reflexes, and no luck at all once the board is set up. Hidden mines are scattered through a grid, and every cell you open reveals a number showing how many mines touch it, counting all eight neighbouring cells. Use those numbers to deduce exactly where the mines must be, flag them, and safely open everything else. Open a mine and the game ends instantly; clear every non-mine cell and you win.
Mobile: tap a cell to dig it. To flag a suspected mine, either long-press the cell for about half a second, or tap the pickaxe button in the HUD to switch into a dedicated flag mode so every subsequent tap plants or removes a flag instead of digging.
Desktop: left-click to dig a cell, right-click to flag it directly without switching modes. You can also move a keyboard cursor around the grid with the arrow keys or WASD, dig the cell under it with Space or Enter, and flag it with F — useful for playing without touching the mouse at all. P or Escape pauses the run, and M mutes sound.
Mines aren't placed on the board until after you make your very first dig — the game waits to see where you click, then scatters mines everywhere except that cell and its immediate neighbours. That guarantees the board always opens into a usable region with real information to work from, so you'll never lose on move one, and you never have to guess blind just to get started. Every loss after that point is a genuine reasoning mistake rather than bad luck on an opening click.
Once you've correctly flagged as many mines around an already-open number as that number states, clicking the number again opens every one of its remaining unflagged neighbours at once — this is the standard "chord" move, and it's how experienced players clear large boards far faster than digging one cell at a time. It is also genuinely unforgiving: the chord doesn't re-check your flags before firing, so if even one of them is on the wrong cell, the chord will detonate a mine exactly as if you'd dug it directly. Chording is worth learning once you're comfortable with basic deduction, but it rewards confidence you've actually earned, not confidence you're guessing at.
Three difficulties are available from the start screen. Easy is a 9×9 grid with 10 mines, a good size for learning the deduction patterns. Medium steps up to 12×16 with 30 mines, and Hard is 14×20 with 55 mines, where a single misread number can cost you a run that took several minutes to build. Your best result on each difficulty is recorded as the fastest winning time, tracked separately per difficulty so an Easy record never gets mixed up with a Hard one — and only a win counts toward it, since losing a board doesn't log a time at all.
The most useful cells on the board are the low numbers next to a lot of already-open space, since a 1 or 2 touching mostly-cleared ground leaves very few possible cells for its mines to occupy. Work outward from those first rather than staring at dense, high-numbered clusters early on. When two numbers overlap on the same unopened cells, comparing what each one implies is often the only way to break a spot that looks like a pure guess — subtracting one constraint from another can pin down a mine, or clear a cell, that neither number alone could resolve. On the larger boards especially, working the edges and corners first tends to open up more free space per dig than probing the middle, since corner cells have fewer neighbours and therefore fewer mines to account for.
The clock starts the moment your first cell opens and runs continuously until you either win or hit a mine, so there's no separate "thinking time" that gets excluded — every second you spend studying the board before your next click counts against your final time exactly the same as a second spent actively digging. That's a deliberate trade-off: it means the fastest recorded times tend to belong to players who've internalized common opening patterns and don't need to pause and re-derive them from scratch on every board, rather than players who simply click quickly without thinking. Because a loss records no time at all, there's also no incentive to rush into an unsafe chord or a low-confidence guess just to protect a partial run — a slower, verified win always beats a fast attempt that ends in a detonation, since only the win goes on the board.