Autoresearch 把 harness 的思想推进到实验系统：定义 objective、logging、rollback、search strategy 和 program。它更像“研究自动化”的作业系统。
适合怎么读
先看本专题首页，再按卡片进入各原始材料；若需要站内全景，可跳到站点地图与模板索引。
本专题包含
14 个原始条目，均保留源文件路径、站内相对链接与外部参考链接。
专题抓手
- 目标函数和边界显式化
- 搜索与回滚机制
- 适合实验、优化、research loop
文档清单
Framework
An autoresearch system is a bounded experimental loop operated by one or more agents under explicit human defined policy.
Ideas
把 autoresearch 放在 harness engineering 下面，不是因为它只是一个小分支， 而是因为它天然依赖 harness：
Autoresearch Manifesto
Autoresearch 不是“让 AI 随便跑一跑实验”。
Philosophy
清晰目标 明确边界 可见状态 可审计结果 可撤销后果
autoresearch
把 autoresearch 视为 harness engineering 的一个子主题 ：
Minimal Program Example
Minimize metric x . Lower is better.
Harness Engineering vs Autoresearch
Autoresearch is best understood as a specialized branch of harness engineering.
Pattern: Logging
Logging converts experiments from isolated events into reusable organizational memory.
Pattern: Objective
Define what the system is trying to improve, under which constraints, and according to which tradeoffs.
Pattern: Rollback
Rollback ensures failed or non beneficial changes do not poison the current best known state.
Pattern: Search Strategy
Define how the system explores the space of possible improvements instead of making arbitrary edits.
Program: Code Optimization Loop
Your goal is to improve the implementation quality of a codebase under a measurable validation harness.
Program Template
Your goal is to improve the target metric of this system.
Program: Research Loop
Your goal is to improve a research target through repeated, bounded experimentation.