This paper presents a system for step-by-step reasoning and planning in policy-aware autonomous agents. Such agents operate in environments governed by policies and norms, where achieving critical goals may require selectively violating policies while accounting for associated penalties. Unlike prior approaches that primarily enforce compliance, our system supports reasoning about trade-offs between policy compliance, goal urgency, and execution cost during planning. The system extends the AAA (observe–diagnose–plan–execute) agent architecture with policy-aware planning capabilities based on declarative reasoning in Answer Set Programming (ASP). At each step, the agent evaluates alternative plans, estimates policy violations and penalties, and selects actions according to a desired level of compliance. The implemented prototype offers an interactive interface through which users can inject observations at different time points and inspect the agent’s reasoning process, including generated plans, projected completion times, potential policy violations, and explanatory diagnoses. By combining planning, diagnosis, and policy-aware reasoning in a unified declarative framework, the system demonstrates how logic-based approaches can support transparent and explainable autonomous decision making in dynamic environments.