Mobile agents are programs that travel through a computer communications network and have a persistent identity and local state. Agents can communicate with other mobile agents or with stationary (server) agents. Agents may travel to become co-resident with information or processing resources, to propagate data, or even to do business on behalf of their users. Possible applications include network management, service infrastructures, massively distributed information systems, and electronic commerce.

First implementations of systems supporting mobile agents already exist. General Magic's Telescript language supports sending and receiving agents from remote machines. Sun Microsystems released the Java language, which can be used in combination Sun's WWW browser. Active electronic mail over the Internet is another example.

The computational metaphor of mobile agents offers unique opportunities for structuring and implementing dynamic and open distributed systems and applications. But, numerous fundamental open technical problems must be addressed before the technology becomes usable:

• Security issues: How can the system and the infrastructure be protected against malicious agents and hosts? How can an agent or site be protected from other agents?
• Agent interaction: How do agents communicate with each other and with the environment? What are good programming constructs for specifying such interaction?
• Agent control policies: How should troops of agents be organised to accomplish a goal? How can system resources be managed and shared among a collection of agents?
• Coping with heterogeneity: What mechanisms are needed for agents to be able to migrate from site to site?

These problems all cluster around two traditional sub-fields of computing: one is programming languages, the other is operating systems. The aim of the workshop therefore was to bring together a diverse group of researchers in these fields, as well as scientists from industry who are interested in agent technology and who are committed to realise such systems. The goal was to shape a research agenda for an increasingly relevant problem area.

The organisers of the workshop want to thank the European Community for supporting this workshop by the Training and Mobility of Researchers Program. The program, which provides financial support for travel costs and accommodation fee in Dagstuhl, enabled that selected young researchers had the possibility to attend this workshop.