Humanagents typically evolve a set of standard routines for carrying out often-repeated tasks. These routines effectively compile knowledgeabout howto carry out sets of interacting tasks without causing harmful interference. Bymodifyingits routines in response to observed failures, an agent can refine enlarge and refine its planning repertoire over time. Weare constructing a software agent, RAFTER,that applies this incremental improvementstrategy to the managementof planning routines for use in the Truckworldsimulator. Currently, weare building in the capacity to makea variety of repairs to the methodused in a routine to coordinate constituent tasks; in the future, RAPTERwill be extended to include methodsfor coordinating tasks fromdifferent routines.