You would not want to perform inbound summarization, and here's why: The first criteria a router uses to determine the best path to a destination is the specificity of the route (the length of the subnet mask). Only in the event that two routes have identical masks will the router compare the Administrative Distance (through which protocol the route was learned). And only if two routes have identical ADs will the metric be compared (how far away the protocol says the destination is).
Therefore, if the router automatically performs summarization for routes it learns from neighbors, that summarization could cause packets to begin using undesirable paths to their destination. Even if you, as the network engineer, really wanted to perform the summarization anyways, it would still be a bad idea because those routes you are summarizing are likely coming from different originators (different AS numbers, operated by different people). Those routes are subject to change if the owners switch to a different ISP, switch circuits, or have a failure scenario. If you manually performed inbound prefix summarization you would be setting yourself up to troubleshoot connectivity issues on a regular basis when things perform badly or break for a period of a few hours at a time. One example would be a route loop where one provider revokes a specific prefix, yet your summarization still points to that provider and traffic goes into a black-hole.
Summarization is a task that only advertisers should perform, not learners. That is why enterprise-grade equipment includes mechanisms to summarize outbound while not including mechanisms to summarize inbound.