In the past, Multi-agent Systems were used in proprietary environments. Nowadays, these systems have been used broadly in open distributed networks, such as e-commerce applications for Internet. An environment such as the Internet cannot be considered a safe place. Thus, multi-agent systems should have security mechanisms, e.g. confidentiality and integrity. The XML Security Specifications are standards that are based on XML and provide security mechanisms. They include: XML Digital Signature for digital signature; XML Encryption for cryptography; XML Key Management Specification for Public Key Infrastructure. Agents may use a FIPA standard called RDF, which is a message content standard in XML language. Using this standard, agents can communicate exchanging XML messages, but these messages are not secure. In this article, we propose a secure communication model for agents based on RDF and the XML security specifications.