Towards Fine-Grained Proof Planning with Critical Agents

Agent Based Command Suggestion Mechanism In order to support users within large proofs, most interactive theorem provers have some limited capability to suggest parameters of choosen commands or even commands themselves. However, those mechanisms generally have the shortcomings that they use in exible sequential computation strategies (i.e., although most parameters in uence the suggestions of each other they are suggested in some order such that parameters choosen earlier in uence those required later, but not vice versa) and do not su ciently use available system resources by computing suggestions inbetween single interaction steps.