rmutt is a utility for generating random text from context-free grammars (a.k.a. recursive transition networks). It's modeled after Andrew C. Bulhak's late, great "dada engine", which he used to write the hilarious Postmodern Thesis Generator. rmutt takes as input a set of user-supplied grammatical rules, each of which represents a set of choices that can be made at a particular level of grammatical description. For instance, a grammar might specify that a "person" can be either "Fred" or "Jennifer". rmutt then makes these choices randomly, resulting in text which conforms to the grammar but is otherwise unpredictable. Grammars in rmutt need not relate in any way to the grammar of any human language. For instance, rmutt can be used to generate text instructions for producing graphics or music. rmutt is similar to the dada engine in that it allows for contextual behavior. In other words, its grammars are not actually context-free. For instance, a choice made in one rule may affect the definition of another rule. Also, textual transformations can be applied to the output of rules.