\b asserts that the current position in the string is a word boundary. \B negates the assertion: it asserts that the current position is not a word boundary. Both are assertions, so unlike other character escapes or character class escapes, \b and \B don't consume any characters.

A word character includes the following:

• Letters (A–Z, a–z), numbers (0–9), and underscore (_).
• If the regex is Unicode-aware and the i flag is set, other Unicode characters that get canonicalized to one of the characters above through case folding.

Word characters are also matched by the \w character class escape.

Out-of-bounds input positions are considered non-word characters. For example, the following are successful matches:

/\ba/.exec("abc");
/c\b/.exec("abc");

/\B /.exec(" abc");
/ \B/.exec("abc ");

The following example detects if a string contains the word "thanks" or "thank you":

function hasThanks(str) {
  return /\b(thanks|thank you)\b/i.test(str);
}

hasThanks("Thanks! You helped me a lot."); // true
hasThanks("Just want to say thank you for all your work."); // true
hasThanks("Thanksgiving is around the corner."); // false

• Assertions guide
• Regular expressions
• Input boundary assertion: ^, $
• Lookahead assertion: (?=...), (?!...)
• Lookbehind assertion: (?<=...), (?<!...)
• Character escape: \n, \u{...}