From Claude Code to Codex — Multi-Agent Coding to the AI Scientist

From external harness to internalized autonomy, and onward to LLM-Wiki-driven research