Harness-RSI README

项目的 README,来自 github.com/zgcagi/Harness-RSI。合并为一个 markdown 文件:README.md。

Harness Self-Evolution: Theory,
Algorithms, Evaluation, Applications, and Impact

📄 Full version · 66 pages

Self-Evolving Agent Harness · arXiv · Project Page · Reading List · Citation · 中文版 · Awesome

Chuyang Wei1,3, Haoxiang Guan1,3, Minghang Zhu1, Kexin Zhou1,3, Jiyan He1,2, Xiren Zhou3, Huanhuan Chen3, Yifei Shen1,2,†

1 Zhongguancun Academy  ·  2 Zhongguancun Institute of Artificial Intelligence
3 University of Science and Technology of China  ·  † Corresponding author

The path of harness self-evolution, from instructions to executable components and runtime

Companion materials for our survey of how execution experience becomes persistent harness state under a fixed base model. The survey covers update mechanisms, evolutionary search, continual PAC learning, evaluation, applications, and societal impact.

PDF downloads: Full report · 66 pages · Short paper · 18 pages.

Reading Guide

Background Related Surveys · Definition · Taxonomy · Precursors

Literature Interpreted · Executable · Theory · Evaluation

Resources Applications & Outlook · Reported Evidence · Contributing · Citation

Definition

A harness $H$ is the software and configuration that governs how a model $M$ operates in an environment $E$. It includes tools, context construction, memory, and execution control. The final evaluator $J$ scores a trajectory and stays outside the candidate’s editable surface; internal critics and provisional verifiers may belong to the mutable harness.

$$ \tau \sim \mathrm{Run}(M,H,E), \qquad y=J(\tau) $$

Harness self-evolution uses a batch of trajectories $\mathcal{B}_k$ to revise the state used by later runs:

$$ H_{k+1}=\mathrm{Update}(H_k,\mathcal{B}_k) $$

The update mechanism remains available for further revisions, and an accepted state is saved and loaded by a later run. Rejected candidates leave the harness unchanged. The analysis holds $M$ fixed; systems that also train weights are read here for their harness-side changes. Offline optimizers and mechanisms with limited persistence are marked in the entries.

Self-evolution is the process of updating persistent harness state; self-improvement is a beneficial change established through evaluation. Harness RSI is the conventional short name used throughout this resource.

Interpreted and executable update targets and their operational loop

Harness components (paper §2)

Runtime responsibilities and mechanisms. An implementation may combine or omit components.

Harness components, runtime responsibilities, and mechanisms

Taxonomy

Interpreted updates revise content read by a fixed host; executable updates revise the implementations it invokes. Evaluation forms a separate branch, while theory cuts across update targets.

Taxonomy and citation map of harness self-evolution

Reference numbers follow the manuscript; source links appear in the reading list below. See also the taxonomy data.

Precursors

1. Interpreted Update Targets

These methods revise instructions, memory, or coordination specifications read by fixed components. Entries note offline optimization and task-local adaptation where applicable.

1.1 Prompts & Instructions

Use candidate scores to propose and select prompt revisions.

Textual-Gradient Critique

Translate errors, critiques, or execution traces into edits to text parameters.

Self-Referential Mutation

Search over the instructions that generate later mutations.

1.2 Memory & Reusable Procedures

Experience Mining

Distill execution trajectories into knowledge retrieved by later runs.

Procedural Skillbooks

Store an explicit procedure that can guide later tasks.

Gated Consolidation

Revise, compress, or replace stored content through a validation or maintenance rule.

1.3 Workflows & Topologies

Fixed-Pipeline Compilation

Optimize prompts and demonstrations while keeping the pipeline structure fixed.

Search over connections and coordination among fixed operators.

Feedback-Driven Workflow Revision

Use execution feedback to revise the workflow specification.

2. Executable Update Targets

These methods revise tools, component implementations, control programs, or plugins invoked by the host.

2.1 Tools & Code Patches

Callable Tool Synthesis

Generate reusable programs exposed through a tool interface.

Evidence-Grounded Component Patching

Modify a declared executable component while the enclosing executor stays fixed.

2.2 Repository & Runtime Logic

Search over the control program; VERSE also revises the optimizer that proposes it.

Repository Revision

Edit a versioned harness implementation and retain candidates for further evolution.

In-Process Patching

Apply changes to a running process; persistence depends on the system’s save and reload mechanism.

2.3 Everything Is a Plugin

Additional reading outside the fixed evidence audit: PluginRSI alternates single-plugin mutation with workflow recomposition in an offline search, retaining useful variants in a shared library. Its updates span role and skill instructions as well as tool and memory implementations. [code]

Contract-Driven Hooks

Bind updates to hooks with explicit interfaces and admission checks.

Microkernel Platforms

Keep a small host fixed and expose replaceable components through plugin interfaces.

3. Theory

The paper develops two perspectives: optimization and generalization. Many reviewed methods search discrete harness state without gradients; scores support selection while traces and critiques guide proposals. Evolutionary paradigms describe shared search structures, and continual PAC analysis studies generalization, retention, and candidate admission.

3.1 Evolutionary Paradigms

The paper maps variation, selection, archive retention, and budget allocation onto EA paradigms. A method may appear in several rows; each mapping states its scope.

Evolutionary paradigms, mapping rationale, and representative harness and skill methods

3.2 PAC Learning & Stability

The continual PAC analysis separates sensitivity to training samples, retention of earlier capabilities, and admission of new candidates. It assumes independent exogenous task draws, task losses in $[0,1]$, and a fixed model and evaluator.

For at most $K$ candidates in each of $T$ rounds, compare each candidate with the incumbent on $m$ fresh tasks independent of its construction:

$$ \widehat{\Delta}_k(H') + \sqrt{\frac{2\ln(KT/\delta)}{m}} < 0 $$

Here $\widehat{\Delta}$ is candidate loss minus incumbent loss. Under the stated assumptions, this gate gives joint confidence $1-\delta$ for accepted candidates on their round’s task mixture. It applies to the candidate actually tested, including any merged edits. Repeated rollouts reduce execution noise but do not add independent tasks.

Stability & Retention

Confidence-Gated Admission

Self-Improvement Limits

3.3 Background Statistical & Optimization Tools

4. Evaluation & Benchmarks

The evaluation framework connects Selection, Activation, Attribution, and Improvement to the evidence supporting each reported comparison.

Four evaluation claims and longitudinal controls

Activation distinguishes an artifact being available, its changed path being reached, and its effect being carried into subsequent execution. See the criteria for evidentiary requirements and coverage.

The Improvement criterion records how reporting tasks relate to those that formed the update. Selection and Attribution assess the reliability and source of the observed gain.

4.1 Claim-Specific Evidence

Selection & Oracle Integrity

Separate reporting evidence from the update process and protect the scorer from candidate edits.

Runtime Activation

Record whether the accepted artifact operated in the runs that produced the result.

Attribution Controls

Separate the effect of the update from execution variation, inference compute, and evaluator changes.

Generalization & Retention

Test on held-out tasks, new settings, or earlier distributions.

4.2 Longitudinal & Dynamic Evidence

Controls for successive updates (paper §6.3)

Successive updates require state and evaluator records, execution variation, inference budgets, and historical task performance. The table connects each control with its confound.

Controls, required records, and confounds in successive-update evaluation

Active-Set Integrity

Track which stored artifacts remain reachable and used after later updates.

Synthesized Boundary Tasks

Construct new tasks while recording overlap with the feedback used to form updates.

Temporal Streams

Use tasks released after a cutoff and account for changes in difficulty and content.

Adversarial Play

Report changes to the agent, task generator, and evaluator separately.

Lifelong Testbeds

Evaluate sequential tasks together with the state carried between them.

5. Reported Evidence

Evidence for the four claims across 93 primary reported comparisons; 2 methods without empirical evaluation are excluded.

Evidence by claim and overall coverage

The most common evidence profile meets Selection and task distance, with partial Activation and Attribution: reports often evaluate on held-out tasks, but less often confirm that accepted updates execute in scored runs or separate their effects from run variation and extra inference compute.

See the full results for comparisons, judgments, and sources, and the criteria for their definitions.

6. Applications and Outlook

The survey examines personal assistance and developer workspaces, enterprise automation, scientific discovery, robotics, and public or clinical services. Harness state can preserve preferences, shared procedures, research know-how, and executable skills. These settings differ in what an update changes, how far it travels, and how its effects can be checked.

Persistent and shareable artifacts also affect work and learning, control over accumulated knowledge, privacy, security, and institutional accountability. Open questions include model–harness–environment co-evolution, optimization over heterogeneous artifacts, compositional program contracts, reusable state, dynamic evaluation, and authorization boundaries. See the application map and societal-impact map, with the full discussion in Sections 7–9 of the paper.

Examples include runtime capability bindings in DSH Everything, clinically prioritized tool admission in MedRSI, and practice-environment design in EnvHarness.

The short paper presents cases on model–harness pairing, plugin reuse, and distributed execution.

Contributing

See CONTRIBUTING.md for adding papers, correcting evidence, and rebuilding the site. Keep English and Chinese entries synchronized.

Bibliographies: full report · short paper.

Citation

Please cite the version you use:

@misc{wei2026harnessrsi,
  title  = {{Harness Self-Evolution: Theory, Algorithms, Evaluation,
             Applications, and Impact}},
  author = {Wei, Chuyang and Guan, Haoxiang and Zhu, Minghang and
            Zhou, Kexin and He, Jiyan and Zhou, Xiren and
            Chen, Huanhuan and Shen, Yifei},
  year   = {2026},
  url    = {https://github.com/zgcagi/Harness-RSI/blob/main/paper/harness-self-evolution.pdf}
}

@misc{wei2026selfevolvingagentharness,
  title         = {{Self-Evolving Agent Harness}},
  author        = {Wei, Chuyang and Guan, Haoxiang and Zhu, Minghang and
                   Zhou, Kexin and He, Jiyan and Zhou, Xiren and
                   Chen, Huanhuan and Shen, Yifei},
  year          = {2026},
  eprint        = {2610.18813},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/pdf/2610.18813}
}

Released under CC BY 4.0. Listed works retain their respective licenses.

Harness-RSI (zgcagi/Harness-RSI)

正在搜索数据库…

这是什么?这是综述《Harness Self-Evolution》的配套阅读列表和资源网站。该综述探讨 LLM 智能体的 harness(提示词、记忆、工作流)如何在底层模型权重保持不变的情况下,根据执行经验进行更新。
这是论文还是精选阅读列表?两者兼有:它将完整的综述论文(两个 PDF)与一份结构化、附带链接说明的阅读列表打包在一起,列表按照论文自身的分类体系组织。
阅读列表是怎么组织的?通过阅读指南依次浏览:背景(综述、定义、分类体系、前期工作),文献部分分为解释型、可执行型、理论型三个分支,然后是评估、应用与展望,以及已报告的证据。
这份列表实际涵盖哪些方法?提示词/指令优化(APE、OPRO、GEPA)、记忆与技能库系统(Reflexion、ACE、ReasoningBank),以及工作流/拓扑结构搜索(DSPy、GPTSwarm、AFlow)。每条条目都标注了发表出处,并附有一句话摘要。
在这个框架里,到底什么算“harness”?指模型在环境中运行所依赖的软件与配置,包括工具、上下文构建、记忆和执行控制;最终的评估器则不在 harness 可编辑的范围之内。
新手应该从哪里开始?建议先阅读“定义”和“分类体系”两节,掌握相关术语;然后根据兴趣选择解释型、可执行型或理论型中的一个文献分支。每条条目都链接到论文、代码或项目主页。
这里的自进化和自我改进有什么区别?自进化指任何更新持久 harness 状态的过程;自我改进则特指经评估确认有益的改动,而非任何更新。
为什么叫“Harness RSI”?Harness RSI 是该项目对 harness 自进化的简称,指将递归自我改进(recursive self-improvement)应用于智能体的软件脚手架,而非其模型权重。
有中文版吗?有。项目在英文资料之外还提供了中文版,因为它被描述为双语阅读列表和项目网站。
除了阅读列表,这份综述还涵盖哪些主题?它涵盖更新机制、进化式搜索、持续 PAC 学习理论、评估方法、应用,以及自进化智能体 harness 的社会影响。
作者
zgcagi
大小
44.5 MB
星标
451
更新
提交Nov 13
时长
1个月
提交 · 90 天