In most practical situations, agents need to continuously improve or repair their plans. In a multiagent system agents also need to coordinate their plans. Consequently, we need methods such that agents in a multiagent system can construct, coordinate, and repair their plans. In this paper we focus on the problem of coordinating plans without exchanging explicit information on dependencies, or having to construct a global set of constraints. Our approach is to combine a propositional plan repair algorithm for each agent with a blackboard that auctions subgoals on behalf of the agents. Both the details of a first construction and some initial experimental results are discussed.