WordTools

Live at masterfool.github.io/WordTools — nothing to install, and it keeps working offline once the page has loaded.

A word finder for Waffle, Wordle, crosswords, and anything else where you know some letters and need the words that fit.

Type the word slot as it currently sits in your puzzle, tap each tile to mark it green / yellow / gray, and the matching words appear as you go. No build step, no server, no dependencies — it is a handful of static files.

Using it

Length sets how many tiles there are, from 3 up to 15. However long the word, the tiles stay on one line — they narrow to fit rather than wrapping.

The tile row is the word as it stands right now. Type the letters — they land gray — then click a tile (or press ↑/↓) to change its colour:

Color	Means	Constraint
⬜ gray	not in the word at all	bans the letter
🟨 yellow	in the word, but not at this position	requires the letter, bans this slot
🟩 green	this letter is at this position	fixes the position

Colours cycle in that order, gray → yellow → green, each press committing a little more. Typing a new letter over a tile resets it to gray.

Duplicate letters work the way they do in Wordle: one yellow E plus one gray E in the same row means "exactly one E" — you do not have to say so separately.

"Not here" sits under the tiles, one box per position. Type letters that cannot go in that slot. Use it for the Waffle case where a letter is gray in the perpendicular word, so you know it is not at the intersection — a yellow tile would be wrong there, because yellow also claims the letter is somewhere in this word. Letters already implied by the tiles show dimmed beneath each box.

The letter panel counts on top of the tiles. That is the whole trick, and it is why there is no min/max fiddling anywhere:

• Also contains X — the word needs another X beyond the ones already on the tiles. Use + for two or more extra.
• No more of X — the tiles show every X there is; no others.

So with SA__Y on the tiles (S green) and N,D gray:

• no more of S keeps the S you placed but rules out a second — SASSY drops out.
• also contains S demands a second S — SASSY becomes the only answer.

With an empty tile row these mean exactly what you would expect: "the word contains X" and "the word has no X".

The A–Z grid is the whole alphabet at a glance. Click a letter to cycle neutral → also contains → no more of. A dot under a letter means the tiles already settled it: green placed, yellow required, gray ruled out.

Contradictory constraints produce an explanation, not a silent empty list.

Results are sorted most common first by default, with a divider before the words that are in the word list but that nobody actually says. Click any word to copy it. The whole query lives in the URL, so you can bookmark or share it.

Running it locally

fetch() and ES modules do not work over file://, so serve the directory:

python -m http.server 8000

• App — http://localhost:8000
• Engine tests — http://localhost:8000/test.html
• UI tests — http://localhost:8000/test-ui.html (drives the real page with synthetic events; run it in a normal browser window)

Nothing to install. No Node, no npm, no bundler.

Layout

index.html            app shell
test.html             engine tests
test-ui.html          interaction tests, driven through a real page
css/style.css
js/engine.js          pure constraint logic — filter, derive, merge, conflicts
js/dictionary.js      loads and indexes the word list
js/state.js           URL hash <-> app state
js/ui.js              all DOM
data/words.txt        196,494 words, 3 to 15 letters
data/common.txt       39,893 words ranked by frequency
tools/build_common.py regenerates data/common.txt (needs network; output is committed)

See CLAUDE.md for the constraint model everything is built around.

Data

data/words.txt is a list of 196,494 words of 3 to 15 letters, lowercase a–z only, one per line. Two-letter words are not in it — they are no use in the games this is for — and the list is curated, so words may be added or removed over time.

data/common.txt ranks 39,893 of those by frequency, derived from hermitdave/FrequencyWords (OpenSubtitles counts, MIT licensed). It is what powers "common first" sorting and the "common words only" filter. If the file is missing the app falls back to A–Z sorting and says so.