Standard agency contracts are enforced by means of Courts since it is assumed that the output results on which payments to the agents are conditioned are fully verifiable. This implies that the Court and the Principal have the same information. This is unrealistic in many contexts since Principal has additional information and better knowledge to evaluate them. Under these conditions, incentive contracts, notwithstanding their (second-best) efficiency, cannot be implemented since there exits a risk of firm's moral hazard. By using repeated game theory and the insights of self-enforcing implicit contracts, we propose a model in which an agency contract is still possible if the firm's reputation can play a role. The model is based on workers with finite life which transmit to the following generations the relevant information about past behavior of the firm and follow a strategy prescribing to exert low effort with opportunistic firm. We point out that the enforcement of an agency contracts without Court's verifiability requires that contractual conditions respect other than the usual participation and incentive compatibility constraints, also an additional "enforcement constraint".