This document introduces the HAHRL-RTS platform for developing hierarchical reinforcement learning agents for real-time strategy games. It discusses the complexity of RTS games and proposes using a hierarchical approach and heuristics to accelerate learning. The hierarchy divides the game into 6 independent sub-strategies like training units, defending, attacking, and gathering resources. Each sub-strategy is modeled as a separate decision process. Heuristics are used to guide learning and feature-based function approximation represents the large state-action space to allow for more efficient reinforcement learning.