Piette, EricDoumont, TomTomDoumont2025-05-142025-05-142025-05-142025https://hdl.handle.net/2078.2/41960Deduction puzzle, with their growing popularity, represent an interesting challenge in the field of artificial intelligence. While a large part of the research has focused on specific games such as Sudoku or Nonogram, little effort has been invested in the development of a general agent capable of solving logic puzzles. This thesis explores the development of such an agent, utilizing the Ludii general game system and the constraint programming paradigm. Our approach employs the XCSP3 formalism and modern solvers for constraint satisfaction problems. It examines the challenges and limitations of integrating these tools, particularly in term of efficiency, scalability for large puzzles and generalization of the results. The findings highlight the feasibility of creating generalist puzzle-solving agents and provide a foundation for future research on the topic.LudiiGeneral game playingArtificial intelligencePuzzleConstraint programmingXCSPtext::thesis::master thesisthesis:49677