A closer look at Clue Grid
Read the runs, not a total
Clue Grid is our browser implementation of a small nonogram-style puzzle. A row clue of 3 means exactly one group of three consecutive filled squares somewhere in that row. A clue of 2 1 means a group of two, then at least one empty square, then a single filled square. It does not mean any three filled squares. Read row clues from left to right and column clues from top to bottom. The same square must satisfy both its row and its column.
Mark one square at a time
Choose Fill square to place a filled mark ■, or Mark empty to write an empty note ×. Tap or click a square, or focus it with Tab and press Enter or Space. Repeating the same action on that square clears that mark. Choosing the other action replaces the mark. Blank means unknown; × is your own deduction, not an automatic error warning. Arrow keys move focus within the board without wrapping across an edge. The control label names the row, column, and current state; its descriptions identify the relevant clues.
Start with what cannot move
A clue of 5 fills an entire five-square line. A clue of 2 2 also occupies the whole available width: two filled squares, one mandatory gap, and two filled squares. By contrast, a single run of 3 has three possible placements in a five-square row. Their middle square is shared by all three placements, so that square is forced, while the other four squares still need evidence from columns. A forced square is a conclusion common to all allowed placements, not a guess about the most attractive pattern.
Use a line check carefully
Check lines reports how many of your currently filled row patterns and column patterns exactly match their clues. Unknown and × squares both count as unfilled for this comparison. A matching row may still put its filled squares in the wrong places for its columns. The check does not identify wrong cells, reveal the hidden pattern, or prove that all your empty notes are justified. The game finishes only when every filled square matches the unique solution; empty notes are optional.
Compare attempts on the same clues
Undo last edit restores the previous state of one cell and reduces Edits by one. Retry this board clears every mark and the edit history but keeps the same clues. Both work after a win. Load teaching example replaces your current puzzle with the fixed position explained below. Load beginner example opens the separate frame-and-center puzzle from the beginner lesson. New puzzle replaces the current puzzle with a different generated position. Loading or retrying returns the cell action to Fill square. A practice URL with the teaching example reloads that example when refreshed; an ordinary random page starts a new puzzle. Closing or refreshing loses your marks, undo history, and edit count.
What the generator guarantees
Random puzzles are built from a five-by-five pattern and accepted only after a row-and-column search finds exactly one solution. Each generated line has at least one filled square. The bounded generator uses a checked teaching pattern or one of its rotations if random attempts do not produce a suitable different puzzle. A unique answer is not a difficulty rating and does not promise that a particular simple deduction will always be enough. The site does not offer a timer, automatic hint, attempt history, or claim that the fewest edits measures reasoning ability.
Try the fixed teaching position
Choose Load teaching example in the game controls. It begins with all cells unknown. Rows R1–R5 run from top to bottom; columns C1–C5 run from left to right. These are its row and column clues:
Rows Columns R1: 1 C1: 1 1 R2: 3 C2: 4 R3: 5 C3: 3 R4: 1 1 C4: 4 R5: 2 2 C5: 1 1
Keep the game open at that exact example while you work through these deductions. A random puzzle has different clues and should not be filled from this answer.
- Row 3 has clue 5, so fill all five squares.
- Row 5 has clue 2 2. Its two runs and one mandatory gap use all five squares. Fill C1, C2, C4, and C5; C3 is empty.
- Columns 2 and 4 each have clue 4, and their bottom squares are filled. Their runs must occupy rows 2–5. Fill their missing squares and mark their row-1 squares empty.
- Row 4 has clue 1 1. Its filled squares at C2 and C4 already provide those two single-square runs. C1, C3, and C5 must remain empty.
- Columns 1 and 5 have clue 1 1, with the required isolated filled squares already at rows 3 and 5. Their row-1 and row-2 squares must be empty; adding a square would extend a run or create another run.
- Row 2 now has empty ends and filled C2 and C4. Its clue 3 requires C3 to join them.
- Row 1 has only C3 available for its clue 1, so fill it. Column 3 now has its run of three in rows 1–3. All clues agree.
Reveal the completed teaching pattern
A filled square is ■ and an empty square is ·. There are fifteen filled squares. You can solve by placing those fifteen fills; writing × in the ten empty squares is optional. This pattern, its clues, and its unique solution are checked independently.
· · ■ · · · ■ ■ ■ · ■ ■ ■ ■ ■ · ■ · ■ · ■ ■ · ■ ■
The game locks the board when the last required fill is placed, then moves keyboard focus to Undo last edit. Undo reopens the previous state so you can inspect that last deduction; Retry starts a clean attempt with the same clues. A complete pattern is a win, while a low Edits value is only a record of your current sequence of actions.
Common questions
How is this different from Light Switch?
Each action changes only the selected square or its note. It does not toggle neighboring cells. The row and column clues constrain where runs of filled squares can go.
Do I have to mark every empty square?
No. Only the filled pattern determines completion. Empty notes are optional aids; an incorrect × can be replaced by selecting Fill square and activating that cell.
Why can one matching row still be wrong?
A run can have the right length in several positions. Its column clues decide which placement is compatible with the rest of the board. Read both directions.
Is every random puzzle uniquely solvable?
The generator counts solutions and accepts a random candidate only when that count is one. Its checked fallback patterns are unique too. This guarantees one answer for the printed clues, not a particular solving technique or difficulty.