The increasing pervasiveness of wide-area distributed computing resources, like computational Grids, has given rise to applications that have inherent problems of complexity, adaptability, dynamism and heterogeneity etc. The emerging concept of autonomic computing holds the key to the self-management of such a multifarious undertaking and provides a way to further build upon this complexity without incurring additional drawbacks. Furthermore, access to Grid services at present is generally limited to devices having substantial computing, network and memory resources whereas most of mobile devices do not have the sufficient capabilities to be either direct clients or services in the Grid environment. The existing middleware platforms like Globus do not fully address mobility, yet extending the potential of the Grid to a wider audience promises increase in its flexibility and productivity. In this paper, we present a component-based autonomic middleware that can handle the complexity of extending the potential of the Grid to a wider mobile audience, by incorporating the features of context-awareness and self-management. We also address the middleware issues of job delegation to a Grid service, support for disconnected operation/offline processing and secure communication.