Trust inversion and its exploitation: A threat model of the Model Context Protocol for agentic AI. J. Nig. Soc. Phys. Sci. 2026;8(4):3748. doi:10.46481/jnsps.2026.3748