Teaching algorithmic thinking, which is one of the fundamental skills in computer science, is a challenging task. Preparatory courses for budding students of computer science at the Vienna University of Technology use a computer program called Theseus to visualise the execution of maze solving algorithms, thus making the discussion of algorithms more tangible. However, Theseus does not explicitly describe the executed algorithms and students are left to wonder what is happening "inside" them while the search for a maze's finish is visualised on screen. This thesis investigates how an interactive learning environment can not only execute but also formulate maze algorithms in order to support the teaching of algorithmic thinking and increase the learning effect for students. After performing an analysis of various didactic options and selecting a suitable approach that is aligned with the constructionist learning theory, the actual implementation of a new online learning environment called Ariadne is described, which integrates a visual programming language for describing algorithms. This allows students and teachers to understand Ariadne's pre-defined maze solving algorithms, to modify them and to even create new algorithms from scratch.