How columnar transposition cipher works
Transposition ciphers change where characters sit, not what they are — every letter of the plaintext appears in the ciphertext, just somewhere else. In the classic columnar method the plaintext is written row by row into a grid as wide as the keyword, and the columns are then read out top to bottom in the alphabetical order of the keyword's letters: key ZEBRAS orders its columns A-B-E-R-S-Z, so the fifth column is read first and the first column last.
This implementation uses the irregular variant with no padding: a final partial row simply leaves the rightmost columns one character short, and short columns are read only as far as they go. That keeps the ciphertext exactly as long as the plaintext — no invented padding characters — and makes the operation reversible for any input length. Spaces and punctuation are transposed like any other character, which historically helped hide word boundaries.
Decryption reverses the process: knowing the keyword gives the same column order and each column's length, the ciphertext is sliced back into columns, the grid is rebuilt, and the plaintext is read row by row. The tool reports the column order it used (key ZEBRAS → order 4, 2, 1, 3, 5, 0) so you can verify the mechanics by hand, and everything runs locally in your browser.