Multi Decoder

General usage

Paste cipher text into the Message box and press Solve. The decoder runs your input through roughly 250 ciphers and surfaces every result, ranked by how word-like the output looks in the selected language.

Most cipher fields are optional. If a cipher needs a keyword, alphabet, pad letter, or number and you didn't supply one, that decoder is skipped and the placeholder text tells you what's missing. Put keywords in Key 1 and custom alphabets in Key 2 when in doubt. If something looks wrong but you're sure of the cipher, try swapping which key holds which value.

Possible words / coordinates

Two summary panels sit at the very top of the results after a Solve. They scan every decoded output and surface the ones that contain recognizable words or valid geographic coordinates, so you don't have to scroll through hundreds of cipher cards by hand. Clicking any match jumps you down to the originating cipher card and briefly highlights it.

Possible words found

A two-pass detector runs over every decoded output:

Pass 1 (token method). When the output has spaces, it is split on whitespace, punctuation is stripped, and each word is checked against the dictionary. If the percentage of real words clears the threshold, the section passes.

Pass 2 (greedy fallback). Many ciphers produce a wall of letters with no spaces. The greedy algorithm slides through the text, always trying the longest possible dictionary word first, and reports the fraction of letters that landed inside a recognized word. Each section lists a sample of the matched words so you can see what the engine picked up.

Word detection preferences

Click the Preferences button on the Possible Words panel to tune:

• Language. 21 dictionaries (English, English-Big, German, Dutch, French, Czech, Swedish, Norwegian, Danish, Finnish, Polish, Portuguese, Spanish, Romanian, Russian, Turkish, Thai, Malay, Indonesian, Hindi, Tagalog). Changing it re-runs detection.
• Threshold (10%-90%, default 70%). Percentage of words that must be recognized for a section to qualify. Lower = more hits, more noise.
• Minimum word length (1-5, default 2). Ignore matches shorter than this. Raise it to cut "to" / "be" / "an" type noise.
• Minimum number of words (1-5, default 3). Floor to prevent a one- or two-word coincidence from showing up.
• Expand all by default. Start with every matching section open, or collapsed so you can drill in one at a time.

Saved per browser.

Possible coordinates found

The coordinate scanner recognizes 11 formats. Each match is grouped by format and clicking the header jumps you down to the cipher card it came from.

UTM, MGRS, OSGB, and Plus Codes are auto-converted to lat/lon for display.

Coordinate detection preferences

Open with the Preferences button on the Possible Coordinates panel. The first three formats (DD, DDM, DMS) are always on; the rest can be toggled off if you know they won't appear in your puzzle, which cuts false positives.

Area filter. Center latitude/longitude plus a ± range (default ±2°). Only coordinates inside the box are surfaced. Accepts flexible input like N35, W097, -97.5, or 35.5.

Minutes filter. Works on the minutes portion only. If the puzzle's final answer should end in minutes near, say, 28 and 31, set those with a ± range and everything outside is hidden. Works alongside or independent of the area filter.

Filter, focus, and jump

Filter hides any section or card whose title doesn't match what you type. Focus / jump lets you pick one cipher (or a whole section) and collapse everything else. Use it when a result list of 250 cards is getting in the way.

On desktop, a sticky "Jump to" bar appears at the top of results; on mobile the same picker docks to the bottom-left corner.

Cryptogram solver (optional)

The checkbox "Also run cryptogram solver (slow)" next to the Solve button enables an automatic substitution-cipher attack on the current message. It is off by default because it lazily pulls in a ~650 KB dictionary and a ~1.6 MB quadgram blob on first use, then runs a hill-climb / simulated-anneal pass per solve.

When enabled, every Solve will also run a 60-restart hill-climb in a dedicated worker. If a plausible plaintext is recovered (at least 20 letters of input), a card lands in the Classical section with the cipher type (Caesar, Atbash, keyword substitution, or general monoalphabetic substitution), a score per character, an optional keyword hint, and up to two runner-up candidates. Polyalphabetic ciphers like Vigenère are not handled here, the Multi Decoder runs its own Vigenère Automatic Solver alongside.

A Refine with locks button opens the full Cryptogram page pre-loaded with the same ciphertext, where you can lock known letters or words and re-run with custom restarts. The toggle preference is shared with the classic Multi Decoder, so flipping it on one page persists to the other.

Share link

The Share Link button copies a URL that re-creates the message and every setting (keys, pad, numbers, alphabet, translate-from/to, language, cryptogram toggle) on the Multi Decoder. Open that link on any device and it will solve the same input the same way.

Found a bug?

If a decoder is producing the wrong output, please reach out with the input, the cipher in question, and what the expected output should be. The more detail, the faster it gets fixed.