Abstract: The intelligent agents can perform the retrieval requests from users via collaboration in a distributed component repository system. The primary problem to solve in retrieval is the selection of data source, i.e., to locate which repositories may possess the target components. The intelligent agents have their own knowledge base. They need the capability to learn new information and update their knowledge base to keep the retrieval results effective. This paper discusses the updatable agents that can incorporate new information into an existing knowledge base. We also propose a method for knowledge description and update based on logical programming. This method can execute the update task following a declarative updating policy. The updatable agents we proposed have clear semantics, and are able to deal with inconsistent information in a relatively intelligent way.