Crossword Online
Choose a compact Mini, a fuller Standard or a longer Deep crossword, then solve the same real game engine used by the specialist Daily. Practice is local-first, deterministic by URL seed and backed by a frozen source bank that can be independently verified.
How the Crossword grid works
Every white square belongs to one Across answer and one Down answer. The number in a starting square is shared when an Across and Down answer begin in the same place. Black squares are separators rather than input cells. In this specialist bank every white square is checked by both directions, so a correct crossing letter always supports two clues at once.
The game keeps an active direction. Clicking a clue jumps to its first unfinished cell. Clicking the same crossing cell again, or pressing Space while it is selected, switches between Across and Down. This avoids a separate mode switch and mirrors the interaction pattern familiar from mature newspaper-style crossword interfaces.
Mini, Standard and Deep
Mini uses a compact 5×5 mask for short sessions and quick clue scanning. Standard expands to 7×7 with more simultaneous crossings and a longer clue list. Deep uses a 9×9 grid for a denser session without turning the page into a tiny desktop-only board. The layout scales to a 350-pixel mobile grid and grows on larger screens.
The three sizes are not separate game engines or duplicate search pages. They are settings inside one Crossword specialist. Practice accepts a deterministic URL seed, which means the same size and seed produce the same source puzzle when a friend, teacher or your future self opens the link again.
Clues and provenance
The frozen bank is assembled from vocabulary already present in the canonical project. When an answer has a plain authored definition in the existing Anagram or Trivia material, the Crossword can use that definition. Common short fill also has a small explicitly authored supplement.
When no honest semantic definition exists in those sources, the clue is transparent wordplay: it asks you to unscramble the answer letters. That is deliberately less clever than inventing a false definition. The release therefore distinguishes definition clues from wordplay clues in the generated asset and keeps the source hashes beside the build summary.
Check, Reveal and clean solves
Check compares only letters you have already entered. Wrong entries are marked, but blank cells stay blank and no answer is inserted. Because Check provides feedback that a clean solve would not have, using it marks the session assisted.
Reveal applies only to the currently active clue. It fills that Across or Down answer and also marks the session assisted. Clean personal bests are reserved for runs completed without either help action. Assisted sessions still count as solved and remain visible in recent local history so the product does not pretend they never happened.
Autosave and privacy
Crossword drafts are saved in browser local storage under the current project schema. Practice drafts are keyed to the exact source puzzle, while Daily drafts are keyed to the calendar date. Reopening the same source restores letters, selected cell, active direction, elapsed active time and whether assistance has already been used.
No account, cloud database or paid puzzle API is required. Clearing site storage clears these records. Migration from V0.110 copies existing local game history into V0.111 while adding an empty Crossword specialist record, leaving the serialized V0.110 store intact for rollback and verification.
How timing works
The timer begins with the first letter rather than page load, so reading the instructions before starting does not inflate the result. When the document becomes hidden, active timing pauses until the page is visible again. This is a convenience measure rather than a claim that browser timing equals laboratory measurement.
Only clean completion time can replace a personal best for the selected size. The site does not convert seconds into an IQ, reading level, attention score or global percentile. Vocabulary exposure, language background, keyboard familiarity, screen size and clue style can all change completion time.
Why the bank is frozen
A frozen source bank makes QA reproducible. The release verifier can examine every grid, entry, crossing, clue, answer and solution rather than trusting whatever a remote service might return today. It also allows deterministic Practice and date-stable Daily play without network access.
The independent verifier does not rely on a green “Solved” message from the user interface. It validates the bank structure directly, runs thousands of deterministic Practice selections, tests perfect and intentionally incorrect completion domains, and checks a full calendar of Daily selections for validity and uniqueness.
What this page does not claim
Crossword results are local puzzle results, not a medical assessment, IQ estimate, literacy diagnosis or population percentile. A puzzle can feel easier or harder because its vocabulary happens to match your background. The useful comparison is your own experience under similar conditions, not a fabricated global ranking.