Anagram solver against a dictionary

Matching is case-insensitive, whitespace in the letter set is ignored, and shorter words are welcome because the solver uses any subset of the available letters. Results are deduplicated and ordered consistently, making the tool useful for word games, vocabulary exercises, puzzle construction, and repeatable application workflows.

Provide letters and a dictionary that fit your puzzle

Start with the complete rack, hand, or pool of letters available in the puzzle. Repeated letters matter: the letters in “letter” permit two uses of “t” and two uses of “e,” while a candidate requiring a third copy will not match. Spaces, tabs, and line breaks in the letter field are ignored, so you may paste a visually grouped set without changing the calculation. Then provide the candidate dictionary as a list of words. The solver does not consult an external word list or guess which vocabulary your game accepts; your dictionary is the authority. That makes the result appropriate for games with house rules, classroom lists, specialist terminology, or a competition’s approved lexicon. Each nonempty dictionary entry is tested independently. A candidate may use every supplied letter or only a subset, so a five-letter rack can produce both five-letter anagrams and shorter valid words. Empty dictionary entries are skipped rather than returned as words. Keep punctuation only when it is genuinely part of both the candidate and available character set, because it is treated as a usable character rather than silently removed.

Understand matching, duplicates, and result order

Matching is case-insensitive and uses normalized text, which means “React” can match letters entered as “TRACE.” The algorithm builds a frequency table for the available characters and then checks each candidate against a fresh copy of those counts. As soon as a word asks for a character that is absent or already exhausted, that word is rejected. This frequency-based approach is important for repeated letters: simply checking whether every distinct character appears would incorrectly accept words that need too many copies. Equivalent dictionary entries that differ only by capitalization or surrounding whitespace are returned once, preserving the spelling of the first accepted entry. The output places longer words first, which is convenient in games where longer plays are normally more valuable, and orders words of equal length alphabetically for stable results. It returns both the ordered words array and a count, so software can display the matches directly and also summarize the result without recounting it. No random choice, network lookup, or changing external dictionary can alter a repeated calculation with identical input.

Use the solver reliably in games and automated workflows

For an interactive puzzle, paste the current letters, supply the relevant dictionary, and scan the longest results first. For automated use, send the same two fields and consume the returned words and count. Because the capability evaluates only the dictionary you provide, you can keep separate lists for different languages, age groups, tournament editions, or product vocabularies without changing the algorithm. It is also useful for puzzle authors: begin with a proposed letter pool, run a curated answer list through the solver, and confirm which intended and unintended solutions are possible. The calculation is deterministic, so generated puzzle checks can be placed in tests and compared over time. Input limits keep browser and API execution predictable: the letter pool accepts up to 256 non-whitespace characters, the dictionary accepts up to 10,000 entries, and each candidate accepts up to 128 non-whitespace characters. If the letter set is empty, missing, or contains only whitespace, the request fails instead of returning a misleading empty result. Invalid dictionary shapes or non-string entries likewise produce a clear input error, helping calling applications distinguish bad data from a legitimate puzzle with zero matches.