Skip to content
Lizely

Word Search

Find eight fully disclosed words across every compass direction with click, touch, or keyboard endpoints and an independently verified unique occurrence for each target.

Privacy: your files never leave your device. All processing happens locally in your browser.

How to use

  1. 1.Choose the first and last cells of a listed word; endpoints must share a row, column, or exact diagonal, and either endpoint order is accepted.
  2. 2.Find BLAZE, CLOUD, DREAM, FLINT, GLOW, HONEY, IVORY, and JADE across all eight compass directions, using the found highlights and word list to track progress.
  3. 3.Press Check words after all eight are marked to earn exactly 1,000 points; incomplete checks are harmless, while two distinct confirmed wrong selections require Restart.

About Word Search

Word Search is a fixed, transparent letter-grid challenge designed for a short deterministic browser session. The board contains eight rows and eight columns, for 64 uppercase cells. Eight target words are hidden: BLAZE, CLOUD, DREAM, FLINT, GLOW, HONEY, IVORY, and JADE. The fixture is original Lizely content rather than a copied newspaper puzzle, dictionary extract, quotation, or rotating external word feed. Reloading never changes the letters, placements, score, or answer.

The grid is fully disclosed by row. From top to bottom it reads TTGSJMIE, NDALBVDQ, IYDEOAPM, LFERJWDU, FJYNEXRE, BYJKOXEI, DUOLCHAD, and YBLAZEMX. Every row contains exactly eight ASCII uppercase letters. The target list is also fixed and visible beside the grid. This makes the puzzle reproducible across browsers and lets tests verify the exact answer instead of accepting a plausible visual match.

All eight compass directions are represented once. BLAZE runs east from index 57 to 61. CLOUD runs west from 52 to 48. DREAM runs south from 30 to 62. FLINT runs north from 32 to 0. GLOW runs southeast from 2 to 29. HONEY runs northwest from 53 to 17. IVORY runs southwest from 6 to 34. JADE runs northeast from 28 to 7. Indexes are zero-based in the disclosed engineering fixture; the interface presents human-readable rows, columns, and letters.

To mark a word, choose its first endpoint and then its other endpoint. The endpoints must share one row, one column, or an exact diagonal. The path may be selected in either order, so choosing the physical end before the start is accepted. This is important for backward words and also makes every forward placement playable in reverse. The selected path must spell one listed target in one direction or the reverse of that target in the other direction.

A single-cell tap selects a start. Tapping the same cell again cancels that start without creating an error. After a valid second endpoint, the complete path is highlighted, the target receives a found state, and the selection clears. Finding an already marked target changes no progress and causes no penalty. Letters remain visible under the highlight, and every cell exposes its row, column, letter, and found state through an accessible label.

Partial strings are rejected. Selecting only BLA from the BLAZE row does not count, because the complete target was not selected. A straight line that spells an absent string is also rejected. Endpoints such as indexes 0 and 10, which form neither a row, column, nor exact diagonal, are rejected before letters are read. These invalid selections set a pending error but do not immediately change the mistake counter or remove already found words.

The Check words action separates ordinary incompleteness from a confirmed wrong attempt. Pressing Check before all targets are found, when there is no pending invalid selection, reports that the puzzle is incomplete and records no mistake. If the most recent completed selection was invalid, Check records one mistake and consumes that pending error. Pressing Check again without making another invalid selection is harmless, so one mistake cannot be charged twice.

The first confirmed wrong check remains repairable. Select any valid target afterward and progress continues from the unchanged grid. A second distinct invalid selection followed by Check deadlocks the run until Restart. This two-error threshold is clearly disclosed Lizely game behavior, not a language standard or traditional word-search scoring rule. It creates a deterministic negative path without turning an exploratory incomplete check into punishment.

Once all eight targets are found, press Check words to complete the run and receive exactly 1,000 points. The score does not depend on speed, selection order, direction, or the number of harmless incomplete checks. It does not award partial points. Completion freezes cell selection, cursor movement, checking, progress, mistakes, and score. Deadlock is frozen in the same way. Restart is the only action that opens a new run.

Restart restores the original 64 letters, logical cursor at index 0, no selected start, no highlighted words, no recent path, no pending invalid selection, zero mistakes, zero score, and no terminal state. Because the fixture never mutates, restart needs no random seed and cannot generate an accidental duplicate or missing answer. The shared double-Escape Boss Mode temporarily replaces the puzzle with a neutral spreadsheet view and restores the same game state when toggled back.

Keyboard play covers the whole puzzle. Arrow keys or W, A, S, and D move the logical cursor across the eight rows and columns. Movement wraps at an edge: moving left from the first column reaches the last column in that row, while moving up from the top row reaches the bottom row in that column. Enter or Space selects the current cell as an endpoint. C runs Check words, and R restarts. Repeated keydown events are ignored so holding a key cannot select two endpoints or confirm two errors.

Mouse and touch use the same native cell buttons. Each target is at least a complete straight path, so no drag precision is required: only two endpoint taps matter. On narrow screens, the 44-pixel cells remain touch friendly inside a local horizontal scroll area. The page itself does not acquire uncontrolled horizontal overflow. Found paths use site design tokens, visible text, and line-through word-list state rather than color alone.

The product scanner walks the grid in east, west, south, north, southeast, northwest, southwest, and northeast order. For each start cell and direction, it reads exactly the target length and rejects any step outside the 8 by 8 boundary. Release tests do not trust that scanner as its own proof. An independent oracle uses separate raw row and direction constants and its own nested loops to enumerate candidate strings.

That independent scan finds each target exactly once in its authored forward orientation. It verifies all eight expected start and end indexes, all row lengths, the uppercase alphabet constraint, and the direction table. Additional tests inspect forward and reversed endpoint strings, exact diagonal cell sequences, the BLA partial boundary, and the non-aligned 0-to-10 boundary. This protects against off-by-one errors and against accidentally treating a bent selection as a diagonal.

State tests cover the complete acceptance path. An incomplete Check produces no error. A wrong TT selection followed by Check records the first mistake, and repeating Check does not record another. A different ND selection followed by Check produces the second mistake and deadlock. A full solve deliberately selects BLAZE backward while selecting the other placements in their authored direction, then verifies all eight found words, exact score 1,000, terminal freezing, cursor wrapping, same-cell cancellation, and clean restart initialization.

Everything runs locally in the browser. No letters, selections, progress, score, or keyboard events are uploaded. No account, API, dictionary service, random generator, or new runtime dependency is required. The fixture contains no personal data and no changing factual reference content. The game remains fast because endpoint validation touches at most eight cells and the independent exhaustive scan runs only in release tests.

This word search is recreational entertainment. It is not a vocabulary exam, reading-level assessment, language proficiency test, educational placement tool, employment screen, health measure, or psychological instrument. The complete fixture and locations are disclosed, repeated play creates familiarity, and the fixed 1,000-point reward is not normed against any population. For an unaided start, inspect the bottom row for the eastward five-letter word, then remember that several other targets travel upward or diagonally.

Methodology & sources

Store eight immutable eight-letter rows and eight target words, with one authored placement in each compass direction. Convert two endpoint indexes to row and column coordinates, accept only horizontal, vertical, or equal-distance diagonal lines, enumerate every intervening cell, and compare the letters with each target and its reverse. Keep invalid selection pending until Check so an ordinary incomplete check remains penalty free; consume one distinct pending error per wrong check, deadlock after two, award exactly 1,000 only after all eight found words are checked, and freeze terminal states. Independently scan every cell in all eight directions from raw fixture constants to prove exactly one forward occurrence per target and the disclosed endpoints without calling product helpers.

Frequently asked questions

Can I select a word backward?
Yes. The game reads the complete straight path and accepts a target in either endpoint order, including horizontal, vertical, and diagonal paths.
Why did my selection not count?
The endpoints may not share a row, column, or exact diagonal, or the complete selected letters may be only part of a target or an absent string.
Does checking an unfinished grid cause a mistake?
No. Check is harmless when there is no new invalid selection. A mistake is recorded only when Check confirms a distinct pending wrong path.
Does the score measure vocabulary or reading ability?
No. The original fixed grid, disclosed answers, exact 1,000 points, and two-error stop are game behavior, not a standardized language, education, employment, health, or psychological assessment.

Related tools

Mini Games guides

View all