This technical note addresses the design of distributed consensus protocols for multi-agent systems with general linear dynamics and directed communication graphs. Existing methods often rely on global information, such as the smallest real part of the nonzero eigenvalues of the Laplacian matrix, which is not feasible for fully distributed control. Instead, this work proposes a distributed adaptive consensus protocol that uses only local information from each agent and its neighbors to achieve leader-follower consensus in any communication graph containing a directed spanning tree with the leader as the root. The protocol is fully distributed and does not require knowledge of the global communication graph structure. The protocol is designed based on the relative states of neighboring agents and incorporates monotonically increasing functions inspired by supply function concepts to provide additional design flexibility. The protocol is shown to achieve leader-follower consensus for any communication graph satisfying the directed spanning tree condition. Furthermore, the protocol is extended to handle multiple leaders, where the followers' states are driven into the convex hull spanned by the leaders' states. A sufficient condition for the existence of the protocol is that each agent is stabilizable. The protocol is validated through a simulation example involving third-order integrators with a directed communication graph. The results demonstrate that the consensus error converges to zero, and the coupling weights converge to finite steady-state values. The proposed method is fully distributed, relying only on local information and is applicable to general directed leader-follower communication graphs. The work extends previous results to handle multiple leaders and provides a new approach for distributed consensus in multi-agent systems.This technical note addresses the design of distributed consensus protocols for multi-agent systems with general linear dynamics and directed communication graphs. Existing methods often rely on global information, such as the smallest real part of the nonzero eigenvalues of the Laplacian matrix, which is not feasible for fully distributed control. Instead, this work proposes a distributed adaptive consensus protocol that uses only local information from each agent and its neighbors to achieve leader-follower consensus in any communication graph containing a directed spanning tree with the leader as the root. The protocol is fully distributed and does not require knowledge of the global communication graph structure. The protocol is designed based on the relative states of neighboring agents and incorporates monotonically increasing functions inspired by supply function concepts to provide additional design flexibility. The protocol is shown to achieve leader-follower consensus for any communication graph satisfying the directed spanning tree condition. Furthermore, the protocol is extended to handle multiple leaders, where the followers' states are driven into the convex hull spanned by the leaders' states. A sufficient condition for the existence of the protocol is that each agent is stabilizable. The protocol is validated through a simulation example involving third-order integrators with a directed communication graph. The results demonstrate that the consensus error converges to zero, and the coupling weights converge to finite steady-state values. The proposed method is fully distributed, relying only on local information and is applicable to general directed leader-follower communication graphs. The work extends previous results to handle multiple leaders and provides a new approach for distributed consensus in multi-agent systems.