learn-claude-code

This is a 2000-word essay about agent architecture that spends 80% of its energy dunking on 'prompt plumbing' and 20% saying 'build harnesses.' The actual code appears to be missing or minimal — the README is truncated mid-word. The core insight (agents need trained models + execution infrastructure) is correct but not novel. The polemical framing against LangChain-style orchestration is fair, but calling everything that isn't RL-trained 'GOFAI' is intellectually lazy. No code examples. No wo...