Took apart Claude Code, then turned it into a free course on harness engineering
The fastest way to learn harness engineering? Build Claude Code from scratch. Why harness engineering? LangChain kept the same model and changed only the harness. Their coding agent went from 53% to 67% on Terminal-Bench, a top coding-agent benchmark. From outside the top 30 to the top 5. Same model. Better harness. Over the past few months, I've taken apart Claude Code, OpenCode, and Pi to see how they work under the hood. And you don't need to ship a coding agent for this to pay off. The same
阅读原始来源来源记录
指标快照
points · 4 votes
2026-10-25 11:00:29.565Z
cumulative since publication
研究分析
来源记录
来源描述:The fastest way to learn harness engineering? Build Claude Code from scratch. Why harness engineering? LangChain kept the same model and changed only the harness. Their coding agent went from 53% to 67% on Terminal-Bench, a top coding-agent benchmark. From outside the top 30 to the top 5. Same model. Better harness. Over the past few months, I've taken apart Claude Code, OpenCode, and Pi to see how they work under the hood. And you don't need to ship a coding agent for this to pay off. The same
现有背景
来源票数: 4 votes (2026年10月25日 UTC; cumulative since publication);前次同口径记录为 4(2026年10月25日 UTC),两个观测值不等于需求增长。 可供比较的近期同类记录:“Stanford CS 329Z: Engineering AI Agents” (Hacker News, 2026年11月13日);“United Airlines blocks AI agents from booking flights” (Hacker News, 2026年11月13日)。主题相近仅用于导航,未核验其是否属于同一需求或独立证据。
研究状态
这条记录尚不能可靠归入主类,未生成商业或性能结论。
缺失证据
需要核验主要事件、受影响对象与原始证据。
下一步
阅读原文,明确它支持什么决策,再选择对应分析框架。