The puzzles differ; the app around them is the same. This page covers what every puzzle gets from it: hints that explain, checking your work, saving and checkpoints, and playing by touch, mouse or keyboard on any size of screen.
Simon Tatham’s manual for the original collection describes his desktop builds, and much of its common features chapter applies here too. Where the two disagree, this page describes the app you are using.
Puzzles you solve by typing, such as Solo and Keen, show an on-screen keypad so they can be played by touch.
If you have a real keyboard and would rather have the screen space, turn the
keypad off in the
Many puzzles use both mouse buttons. When a puzzle’s help says “right-click” and you are playing by touch, you have four ways to do it:
Tap again. In many puzzles, tapping cycles a square: the first tap does what a left-click does, the second what a right-click does, and a third (in puzzles that have a neutral state) clears it.
Long press. Holding a finger down counts as a right-click. To right-drag, keep holding and move.
Two-finger tap. A second finger, anywhere on the screen, turns a touch into a right-click. The fingers need not land together: put the first where you want to click, then tap and release the second. To right-drag, lift the second finger and move the first.
The /
toggle. A button that decides what a plain tap means. It is hidden by
default; turn it on in the
Tapping again is always available where a puzzle supports it. The others can be
tuned or turned off in the
In puzzles that have no use for a right-click, these gestures are switched off — Cube, Fifteen, Filling, Flip, Flood, Pegs and Sokoban. There, holding your finger still simply presses; you can rest a finger on a peg while you decide where to jump it, and the drag still works when you move.
Browsers don’t let a page see a drag made with the right mouse button, so where a puzzle’s help says “right-drag”, hold Ctrl and drag with the left button instead. A plain right-click works as usual.
Many puzzles here have a Next hint button, high in the panel beside the board (or in the bar along the bottom, on a phone). It doesn’t simply reveal a move. It shows you the next move and explains why that move is forced, so that what you take away is a technique you can use again by yourself rather than one square you didn’t work out.
Hint works in two beats:
Press once to see it. The board marks what the hint is talking about and the explanation appears directly under the button. Nothing has been played — the move is still yours to make, and you’re free to make a different one.
Press again to play it. A second press, with nothing done in between, plays that one step in slow motion and stops there. Ask again for the next.
Anything you do in between — a move of your own, undo, redo — puts the hint back to the first beat, so a second press always shows before it plays.
Auto Hint, beside the Hint button, does this continuously at about a second a step, waiting for each move’s animation in the puzzles that have one. pauses it, and it stops by itself when the puzzle is solved or when the hint runs out of things to say.
A hint marks the board, and the marks mean the same things in every puzzle, so they’re worth learning once.
A ring, drawn around the contents of a square, is what the hint is acting on — the square, edge or line that the deduction forces. It is drawn around what’s there rather than over it, so anything you’ve already written stays readable underneath.
An outline around a row, region or area is the evidence — what the hint is reasoning from, as opposed to what it concludes. Where a puzzle’s evidence is squares that are empty by nature, this is a soft shade rather than an outline, since there’s nothing there to read through it.
Small numbers in the corners of squares give the order the steps fall in, when a hint walks a chain of consequences. They’re numbers rather than arrows on purpose: they tell you which order to read the chain in, and stop short of claiming that each square is the one that forces the next.
Some puzzles need to point at two kinds of evidence at once — a filled square and an empty one, say — and mark them differently from each other. The explanation always names what it’s pointing at, so you never have to go by color alone.
Two quite different things stop a hint, and they call for opposite responses, so it’s worth knowing which one you’re looking at.
There’s a mistake on the board. A hint won’t reason from a position that already contradicts the clues, because everything it deduced from there would be wrong too. It refuses, and highlights the squares that clash — the same highlighting Checking your work uses. Fix those, then ask again.
Deduction has run out. Nothing further follows from what’s on the board. The puzzle is still solvable; there just isn’t a next step that can be explained, which is what an Unreasonable puzzle is for. Save your position, try something, and come back if it doesn’t work out — or take the answer from Show solution…, at the foot of the Help me play group.
The hint says which of the two has happened. It never guesses on your behalf and then presents the guess as a deduction: a guess that happens to come off is not a technique, and teaching you one would be the point.
Not every puzzle has a Next hint button. A puzzle has one where it’s solved by reasoning and the game can put that reasoning into words. Where the challenge is dexterity, search or luck instead, there’s no technique to teach and no button to press.
Where a puzzle has a single provable answer, this app can tell you whether what you’ve entered so far contradicts it.
Check & save — beside the board, in the bottom bar on a phone, and on Ctrl/Cmd+S — checks the board and then saves your position if it’s sound. If it isn’t, it tells you how many squares are wrong, highlights them, and doesn’t save; the position you saved earlier is left exactly as it was, so a check you fail can never cost you the one you passed. Return to that saved position at any time with Back to last save, directly beneath it.
In a puzzle that can’t check itself the button reads the same and simply saves — one name for the save in every puzzle.
There is also a quieter Check without saving, low in the Help me play group. It highlights and counts the mistakes exactly as above but writes nothing, which is what you want when you already saved a position you mean to keep: the save slot holds one position per puzzle, so checking-and-saving would replace it.
Two things worth knowing about what the highlighting claims:
There is one saved position per puzzle. It’s separate from the numbered checkpoints in the history panel, which are a different feature: those live in your move history and you can hold several at once.
Many puzzles offer a range of difficulty levels in the type menu. Nearly all of those levels promise the same thing, and one of them promises something different.
On every level except one named Unreasonable, the puzzle can be finished by reasoning alone. You may not spot the next step, and it may be a hard step to spot, but there is always one there to be found, and you never need to try something out to see whether it works.
Unreasonable withdraws exactly that promise. Such a board may reach a position where no further step follows from what you can see, and the only way on is to pick one of the possibilities, play it out, and be ready to take it back. That’s not a flaw in the board; it’s what the name is telling you in advance. It’s also the only name that’s allowed to mean it — if a level is called something else, it doesn’t require this of you.
Three things make trying-and-taking-back cheap:
The hint knows about all this. On an Unreasonable board it will tell you when deduction has run out rather than making a choice for you and calling it a deduction.
In puzzles where you write small pencil marks into a square to keep track of what could go there, the button writes them for you: every square that has no marks yet gets everything that could still go in it. That gives you a full set to eliminate down from instead of a blank grid to fill in by hand.
It never undoes your own thinking. A square you have already narrowed by hand is left alone — the button only fills in squares that are empty of marks. And pressing it a second time doesn’t start over: in most of these puzzles it crosses out the candidates that the board has since ruled out, so repeated presses narrow rather than reset.
On a keyboard this is also the M key. The Fill all pencil marks button is the only way to reach it on a touchscreen.
It’s a move like any other, so undoes it.
Some puzzles are about placing a known set of pieces, where the question you keep asking is which ones haven’t I placed yet. Those get a reference panel: a checklist of the whole set, marking off each piece you’ve placed and flagging any you’ve placed more than once.
Picking an item in the panel highlights where on the board it could still go. Closing the panel deliberately keeps that highlight, because on a small screen the natural order is to pick your piece, close the panel to see the board properly, and then place it. Press Esc, or pick the same item again, to clear it.
A checkpoint marks a position in your move history so you can come back to it: a shortcut for a long run of Undo. They earn their keep on an Unreasonable board, where you may have to try a line and take it back.
You can hold several checkpoints at once. If you want a single position to return to, and the board checked before it is kept, that is the separate saved position.
To set one, open the history panel (beside undo and redo) and choose Save checkpoint, then carry on. If the line doesn’t work out, pick the checkpoint in the panel to rewind to it.
Back at a checkpoint, undo and redo still walk the history around it. Your first new move, though, discards everything after that point, later checkpoints included; if you change your mind before making one, Last move in the panel redoes all the way to the end. A checkpoint you no longer want goes with its delete button.
Checkpoints are experimental and may change shape.
Each puzzle keeps your game in progress and resumes it when you come back, whether you closed the tab, wandered off, or hit a bug. Starting a new game, or choosing another variation or difficulty in the type menu, replaces it.
On the home screen, a puzzle with a game in progress wears a triangle on its icon.
To discard every game in progress, open the
Share in More… (or
This specific game: a link to the game you are playing, as it was dealt. It doesn’t carry your progress; for that, export a save file or copy as text.
This puzzle type: a link to the current type, meaning the size, difficulty and any other options. Opening it deals a new random game of that type, which makes it the way to pass on a custom type.
Copy as text: the board as plain characters, for pasting into a forum post or a message. Use a fixed-width font (“format as code”). Not every puzzle can do this, and some of the renderings take a little imagination.
Game ID and random seed: for use with other apps that play the same
collection. Simon Tatham’s manual explains the format under
Specifying games with the game ID. To open an ID or seed you
were given, use
For puzzles that exist on Simon Tatham’s website, the share dialog also links the same game there, which is handy for comparing behavior when something looks wrong.
Save game in More… keeps the whole game, undo history and checkpoints included, and Load game brings it back. Saved games stay on your device, in your browser’s storage.
Export… in the save dialog writes a file you can move to another app that plays the collection, or attach to a bug report; Import… in the load dialog reads one in. (Exported files leave checkpoints behind for now.)
To delete every saved game, open the