About this tile word finder
A small, fast, dependency-free tool for one job: turning a rack of letters into every playable word, ranked by score. No account, no upload, and nothing you type is ever recorded.
Why it exists
A stuck rack is one of the most common moments in any tile word game: seven letters that do not obviously spell anything, while the clock or your patience runs down. Checking every possible combination by hand is slow and error-prone, and most existing finders either assume one specific game’s scoring, ignore blank tiles entirely, or bury the actual point value under ads and clutter.
This tool does the one job properly: it takes your rack (blanks included), checks it against a large dictionary, and shows you every valid result with its exact score under either of the two most common tile-value systems — instantly, and without sending your rack anywhere.
How the matching and scoring are implemented
The rack is read into a count of each real letter plus a separate count of blank tiles. For a candidate word to be playable, every letter it needs must come from either the rack’s own supply of that letter or a blank standing in for it:
Real letters are always used before blanks for the same letter, because that is never worse for the score — a real letter's point value is always zero or more, so preferring it over a blank can only help or tie, never hurt. This also means the finder never has to guess or offer a choice: for a given rack and word, there is exactly one correct score.
The scoring tables themselves are two separate 26-letter point maps — classic and popular-mobile — described in full on the tool page. Switching the toggle re-scores the same set of matched words against the other table; it does not re-run the rack match, since which words are playable never depends on which system you are scoring with.
How it was verified
Before this page shipped, the scoring and matching functions were run against a set of known-answer cases:
| Case | Expected | Result |
|---|---|---|
| QUIZ, classic system | Q10+U1+I1+Z10 = 22 | match |
| QUIZ, popular-mobile system | Q10+U2+I1+Z10 = 23 | match |
| CAT with a blank standing in for C | 0 + A1 + T1 = 2, not 5 | match |
| Rack CAT against a small list | Returns CAT, ACT, AT — not DOG or CATS | match |
| Word longer than rack + blanks | Correctly rejected as unplayable | match |
| Duplicate entries in the source word list | Collapse to one result each | match |
The contains, starts-with, ends-with and length filters were each checked against a small fixed word list to confirm they include exactly the words that should match and exclude everything else, and the full 26-letter tables were checked against every letter for both systems before shipping.
How the site is built
- Static HTML, one small stylesheet, one small script pair. No framework, no build step, no bundler.
- The finder needs nothing from the network beyond the page itself. The dictionary is a plain word list shipped with the site, not fetched from an external service; the favicon is an inline SVG data URI. Once a page has loaded, searching your rack works offline.
- System font stack. Your operating system's UI font, so nothing has to download and nothing shifts as it loads.
- Nothing you type is transmitted. Matching and scoring happen entirely in your browser against the word list already on the page; your rack is never uploaded to anything.
- Dark mode follows your system setting via
prefers-color-scheme.
Scope and honesty about limits
This tool implements two widely used tile-scoring systems and matches against a large general-purpose English word list. It does not implement any single game's exact proprietary dictionary, it does not know anything about a board (bonus squares, adjacency, what is already placed), and it does not calculate the actual score a play would earn once multipliers and bonuses are applied — only the flat tile value of the letters themselves. See the dictionary note and strategy note on the tool page for the full detail on both limits.