Methodology
Transparent matching, stated limits.
We explain what the tool does, what it does not do, and how to verify a result.
1. Normalize the input
Letters are converted to lowercase and spaces are ignored. ? or * counts as one blank. Other characters and racks outside 2–15 tiles produce a visible error; input is never silently truncated.
2. Count letters
The rack and each candidate are stored as letter counts. A candidate passes when every required count is available; blanks cover a limited number of missing counts.
3. Apply search mode
Subset mode accepts words within the selected length range. Exact mode also requires the candidate length to equal the rack length.
4. Rank results
Longer words appear first, followed by a familiar English tile-value score, then alphabetic order. Ranking is deterministic and not presented as artificial intelligence.
5. State the limits
ENABLE includes uncommon, dated and potentially sensitive words. It is not a substitute for the current official lexicon of a tournament, app, school, or publisher. Scores reflect the actual rack: printed tiles contribute their English tile values and blanks contribute zero. Board multipliers, cross-words and bonuses are excluded.
Dictionary source and coverage
The bundled ENABLE list is distributed as public-domain word data. We retain the upstream file and match unique lowercase a–z entries of 2–15 letters. Read the source record, checksum and licensing references. The file downloads once per page session when you first search. An error is shown if it cannot load; no partial list is silently substituted.
Privacy
The search runs locally in the browser. The letters typed into the tool are not sent to our server. Standard hosting and advertising services may still process technical request data as explained in our Privacy Policy.