Guard Post

A small board, six squares across and five down, with nine walls scattered over it. Put four posts out so that every open square is watched by at least one of them. A post watches its own square, then straight along its row and straight down its column until a wall stops the line — never diagonally, and never through a wall. Four is not a suggestion: every board is checked before it is dealt, and a board is thrown away unless three posts provably cannot cover it and exactly one arrangement of four can. So there is always an answer, there is only ever one, and no lazier answer is hiding anywhere.

The board does the counting for you. Every square nobody is watching yet shows how many of the open squares could ever watch it, and that number is the whole game: a square only two places can watch forces a post into one of those two, and once that post is fixed the next count shrinks. The cursor starts on the smallest number on the board, because that is where the reasoning starts.

It runs entirely on your device: 0 network requests, works offline. No ads, no account, nothing to buy, and no betting, money or prizes of any kind.

1. Focus the board — click it, or press Tab until it is outlined.
2. Move the cursor with the arrow keys, or W A S D. It starts on the tightest square, the one fewest places can watch.
3. Put a post down with Enter or Space. It appears as ▲ and immediately watches its own row and column out to the first wall.
4. Take it back by pressing Enter on it again. That costs nothing, so changing your mind is free — only putting a post down spends one of your placements.
5. Or just click any square to do the same thing.
6. Lift them all with Esc, or the button under the board, when you want to start the arrangement again. Also free.
7. Read the numbers. A blind square shows ◌ and the number of open squares that could watch it. A watched square shows ✓. A wall shows ▨ and stops any line that reaches it.
8. Watch the two counters. "Unwatched" is what is left to solve; "Placements" is what you have spent out of nine. Four of those nine are the answer, so you can afford to be wrong five times before the round ends.

How do you cover every square of a grid with the fewest guards?

Start with the square that has the fewest ways of being watched, never with the square that looks most important. If some square can only be watched from two places, one of those two must hold a post — you do not know which yet, but you know the pair, and that is already a fact you can carry. Fix the posts that are forced, cross off everything they now watch, and recount: the squares that are left will usually have smaller counts than they did, and a second post becomes forced. That is the whole method, and it is why this game prints the count on every blind square instead of leaving you to work it out by eye. The counting is tedious rather than interesting; the decision is the game. The trap to avoid is the opposite habit — putting a post on the square with the longest sight lines because it lights up the most at once. A greedy post often watches a great deal and still leaves a corner that nothing else can reach, and on a board where four posts is the proven minimum there is no spare post to fix it with.

Why is the number of posts fixed at four?

Because "place as many as you need" is a mopping-up exercise with no wrong answer, and this is meant to be a deduction. The four is not a designer's guess either: before a board is dealt, a search runs over every arrangement of one, two and three posts on the open squares and confirms that none of them covers the board, then over every arrangement of four and confirms exactly one does. Fixing the count at the proven minimum is what makes every post carry weight — with a spare post you could patch a mistake, and with a spare post the counting argument stops mattering.

What happens when I run out of placements?

The round ends and the answer appears, marked △ on the four squares that were the only set that works. You get nine placements for a puzzle that needs four, so five wrong ideas are affordable; taking a post back off never costs anything, and neither does lifting them all with Esc, so you can rearrange as much as you like — the budget only counts posts going down. Ending the round rather than letting you grind through every arrangement is deliberate: on a board this size you could eventually stumble onto the answer, and stumbling teaches nothing.

Do posts block each other's view?

No. A sight line is stopped only by a wall or by the edge of the board, so two posts can happily watch each other and a post standing in another post's row does not shorten it. That keeps the rule simple enough to hold in your head, and it is also what makes the board's numbers trustworthy: because what a square can see depends on the walls alone, the count printed on a blind square is a fixed fact about the board rather than something that shifts each time you move a post.

Is there any money, betting or in-app purchase in this?

None, and there never will be. There is no currency, nothing to buy, no lives to refill, and nothing to win or lose but the puzzle. The score comes from the board you finished and how few placements it took, and it stays on your device.

Does it work offline, and can I play it with a keyboard or a screen reader?

Yes to all three. After the page has loaded once, the board, the rules and the search that verified them all live on your device, and nothing is sent anywhere while you play. Arrows or W A S D move, Enter or Space places and lifts, Esc clears, and every square announces itself in words — "a post", "watched", "unwatched, 3 squares could watch it", "wall". Nothing depends on telling colours apart: a post, a watched square, a blind square, a wall and a missed answer each carry their own glyph.

One board at a time, thirty squares with nine of them walled, four posts and nine placements. There is no timer, no undo history and no hint button — taking a post back off is free, so an undo would only repeat what Enter already does. The "gentle", "steady" and "tricky" rating is the number of ways to place four posts and leave exactly one square blind, counted exactly rather than estimated; a board with many of those feels cruel because "nearly finished" keeps being wrong. The numbers on the blind squares are a fact about the board, not a hint about the answer: they tell you which squares are most constrained, and there is no board where following the smallest number blindly is guaranteed to work.

Grid Clues is the same habit on a different shape — a constraint that names a handful of possibilities and the reasoning that narrows them. Gap Logic works the other way round, giving you the relationships and asking for the arrangement, and Grid Paint asks what a square's remaining options cost you rather than what they buy.

Bookmark this page (Ctrl+D, or ⌘D on Mac) or install the app — it works offline the next time you need it.