Classic security systems use a trust model centered in the authentication procedure, which depends on a naming service. Even when using a Public Key Infrastructure as X.509, such systems are not easily scalable and can become single failure points or performance bottlenecks. Newer systems, with trust paradigm focused on the client and based on authorization chains, as SDSI/SPKI, are much more scalable. However, they offer some difficulty on locating the chain linking the client to a given server. This paper defines extensions to the SDSI/SPKI authorization and authentication model, which allow the client to build new chains in order to link it to a server when the corresponding path does not exist.