Automatic planning is a full-fledged discipline in artificial intelligence. It allows describing and solving planning problems in many fields such as robotics, project management, internet navigation, crisis management, logistics and games. The PDDL language, planners and validators associated with this language allow the description, resolution and validation of planning problems. In this paper, we recommend a formal process based on the PDDL language that favors the obtaining of reliable PDDL descriptions. Our process is successfully experimented on a set of representative case studies like the Hanoï Towers and the Sliding puzzle game.