Chevrotain can be used to build parsers for LL(K) Grammars. This means that the number of lookahead tokens needed to disambiguate two alternatives must be a fixed number and known in advance.
For example given the grammar
statement: A B C | A B D | A B E
Chevrotain will look three tokens ahead to decide between the two alternatives.
But given the following grammar
statement: longRule B | longRule C | longRule D longRule: A+
Chevrotain will throw a an error during the parser initialization in this case. This is because there is no fixed number of tokens we can use to choose between the alternatives that is due to a potentially infinite number of "A" tokens that can appear before the "B" - "C" tokens.