It is shown that the known models of knowledge representation and processing do not allow constructing an intelligent problem solver for autonomous mobile intelligent agents capable of performing complex tasks in a priori uundescribed unstable conditions of a problematic environment. To solve this topical problem, the article proposes standard constructions of a knowledge representation model, without reference to a specific subject area, based on polyvariable conditionally dependent predicates. The structure of this type of predicates is given and the conditions are determined under which, as a result of the valuation of variables included in them, true statements are obtained that characterize the necessary conditions for achieving behavioral sub goals and goals in the current situation of an unstable problematic environment. The standard and different in purpose elements of knowledge representation model for autonomous intelligent agents have been developed, which make it possible to form programs of purposeful activity of different complexity associated with the implementation of the formulated task. It is noted that further development of the results obtained in this work is associated with the formalization of mental acts and the development of knowledge processing tools for constructing automatic planning algorithms of the purposeful behavior of autonomous mobile intelligent agents in unstable underdetermined conditions.