Leaked Claude Code Analysis: Just 1.6% Is AI Decision Logic

Researchers at UCL and MBZUAI dissected the source code of Anthropic's AI coding agent "Claude Code" — leaked on March 31, 2026 — and concluded that only about 1.6% of the code implements AI decision logic, while the remaining 98.4% is operational infrastructure such as permission management and tool control. The findings were published April 14 as the paper "Dive into Claude Code: The Design Space of Today's and Future AI Agent Systems" (arXiv:2604.14228).