Extending Classical Planning to the Multi-agent Case: A Game-Theoretic Approach