When agents roam, who can you trust?

Roaming agents threaten a server with theft of assets, system resources and reputation. Generally ignored is the reciprocal threat of server against agent. Not only might a server steal the assets, resources, and reputation of an agent, but it might actually kill the agent. The central security concern for systems that allow agents to roam is how to establish trust, and how to limit risk for both the server and the agent. The travels of the agent are key to establishing trust, and trust determines the level of risk.