001tmf/harness-forge

将Claude代码变成自己的元架构——围绕固定模型原生演进脚手架。

-555)

束缚锻造是一项Claude代码技能，是端到端运行的 利用优化循环 — 提出→分数→保持帕累托最佳→重复——以改进 围绕固定模型编写代码：其内存、检索、上下文构建、摘要、提示 模板和工具选择逻辑。模型从不变;脚手架变得更好。

它是该方法的原生重实现 元束：模型束的端到端优化 （李、奈尔、张、李、哈塔布和芬恩，2026年）。原文 参考回收船 ~1,260 条 Python （claude_wrapper.py + meta_harness.py），其任务是驾驶无头的克劳德：生成一个 会话，解析输出，跟踪工具调用，记录一切，循环。在Claude代码里，那个 运行时已经作为一流工具存在。 因此 Harness Forge 只保留不可约域 逻辑——廉价的配乐器——并以本土编曲的形式表达整个外部循环。整体 搜索线数由~1,260变成~75行。

---

一张照片里的理念

提议者是突变算符。

Harness Forge is a Claude Code skill that runs an end-to-end harness-optimization loop — propose → score → keep the Pareto-best → repeat — to improve the code around a fixed model: its memory, retrieval, context construction, summarization, prompt templates, and tool-selection logic. The model never changes; the scaffolding gets better.

It is a native reimplementation of the method in Meta-Harness: End-to-End Optimization of Model Harnesses (Lee, Nair, Zhang, Lee, Khattab & Finn, 2026). The original reference repo ships ~1,260 lines of Python (claude_wrapper.py + meta_harness.py) whose job is to drive a headless Claude: spawn a session, parse its output, track tool calls, log everything, loop. Inside Claude Code, that runtime already exists as first-class tools. So Harness Forge keeps only the irreducible domain logic — a cheap scorer — and expresses the entire outer loop as native orchestration. The whole search becomes ~75 lines instead of ~1,260.

---

The idea in one picture

seed the frontier with the incumbent harness (the thing to beat)
repeat:
    PROPOSE   k candidate harness variants     ← parallel proposer agents write code
    VALIDATE  each imports / type-checks
    SCORE     each on a held-out-protected eval ← a $0, deterministic scorer
    FRONTIER  Pareto-merge: quality up, cost down, floor-respecting
final: score the frontier once on the untouched test split

The proposer is the mutation operator. The frontier is the search memory. The model is frozen throughout — which is exactly why this fits a fixed / off-the-shelf-API deployment, where you can’t change the weights and the gain has to come from the harness.

The paper’s headline result was +7.7 accuracy points at ~4× fewer context tokens on text classification — a pure harness-side win. Harness Forge reproduces that shape of result natively.

---

Why native?

claude_wrapper.py is a hand-rolled agent runtime. Claude Code is an agent runtime. So every orchestration piece has a native equivalent, and the Python driver becomes redundant: