Show HN: Andon – Toyota Production System for LLM Coding Agents
TL;DR
LLM coding agents (Claude Code, Codex, etc.) have structural weaknesses: blind retry loops, volatile learning, silent spec drift, and gate gaming.
Key Points
- These aren't bugs in specific models — they're properties of goal-optimizing systems.
- id=47304539 Points: 2 # Comments: 0
Nauti's Take
Applying Toyota's factory-floor wisdom to LLM agents is genuinely clever — blind retry loops and silent spec drift are real pain points. But with 2 upvotes and zero comments, the jury's still out on whether this solves the problem or just wraps it in Lean jargon.
Sources