Add README (EN/ZH) and CC-BY-4.0 license

- Add bilingual README with HuggingFace YAML metadata
- Fix Schema First citation accuracy (pilot study findings)
- Add CC-BY-4.0 license file

README.md

@@ -1,3 +1,547 @@
----
-license: cc-by-4.0
----
+---
+license: cc-by-4.0
+language:
+  - en
+  - zh
+tags:
+  - survey
+  - llm-agents
+  - agent-harness
+  - agent-frameworks
+  - harness-engineering
+  - evaluation
+  - agent-memory
+  - multi-agent-systems
+pretty_name: "Agent Harness for Large Language Model Agents: A Survey"
+size_categories:
+  - n<1K
+---
+
+[English](README.md) | [中文](README_zh.md)
+
+# Agent Harness for Large Language Model Agents: A Survey
+
+<p align="center">
+  <a href="https://github.com/Gloriaameng/LLM-Agent-Harness-Survey/stargazers"><img src="https://img.shields.io/github/stars/Gloriaameng/LLM-Agent-Harness-Survey?style=social" alt="GitHub Stars"></a>
+  <a href="LICENSE"><img src="https://img.shields.io/badge/License-CC--BY--4.0-blue.svg" alt="License"></a>
+  <img src="https://img.shields.io/badge/Papers-110%2B-green" alt="Papers">
+  <img src="https://img.shields.io/badge/Version-v2-orange" alt="Version">
+</p>
+
+<p align="center">
+  <img src="assets/architecture_diagram.png" width="720" alt="H=(E,T,C,S,L,V) Six-Component Architecture"/>
+</p>
+
+> ⭐ **This survey is actively maintained. If you find it useful, please star the repo to stay updated and help others find it.**
+
+---
+
+> **The agent execution harness — not the model — is the primary determinant of agent reliability at scale.**  
+> This survey formalizes the harness as a first-class architectural object **H = (E, T, C, S, L, V)**, surveys 110+ papers, blogs and reports across 23 systems, and maps 9 open technical challenges.  
+> 📄 **[Read the Paper](#)** (coming soon)  
+> ✉️ Corrections & suggestions: gloriamenng@gmail.com; chenliyi@xiaohongshu.com
+
+If you find this survey useful, please cite:
+
+```bibtex
+@misc{meng2026agentharness,
+  title   = {Agent Harness for Large Language Model Agents: A Survey},
+  author  = {Meng, Qianyu and Wang, Yanan and Chen, Liyi and Wang, Qimeng and
+             Lu, Chengqiang and Wu, Wei and Gao, Yan and Wu, Yi and Hu, Yao},
+  year    = {2026},
+  url     = {https://github.com/Gloriaameng/LLM-Agent-Harness-Survey},
+  note    = {Work in progress}
+}
+```
+
+---
+
+## 🆕 News & Updates
+
+- **[2026-04-03]** Initial release
+- **[2026-04-07]** Repo updated
+
+---
+
+## Table of Contents
+
+- [Overview](#overview)
+- [Historical Timeline](#historical-timeline)
+- [Harness Completeness Matrix](#harness-completeness-matrix)
+- [Paper List](#paper-list)
+  - [Historical Lineages](#historical-lineages)
+  - [Harness Taxonomy](#harness-taxonomy)
+  - [Technical Challenges](#technical-challenges)
+    - [Security & Sandboxing](#security--sandboxing)
+    - [Evaluation & Benchmarking](#evaluation--benchmarking)
+    - [Protocol Standardization](#protocol-standardization)
+    - [Runtime Context Management](#runtime-context-management)
+    - [Tool Use & Registry](#tool-use--registry)
+    - [Memory Architecture](#memory-architecture)
+    - [Planning & Reasoning](#planning--reasoning)
+    - [Multi-Agent Coordination](#multi-agent-coordination)
+    - [Compute Economics](#compute-economics)
+  - [Emerging Topics](#emerging-topics)
+  - [Future Directions](#future-directions)
+- [Citation](#citation)
+- [Update Log](#update-log)
+
+---
+
+## Overview
+
+LLM agents are increasingly deployed in agentic settings where they autonomously plan, use tools, and act in multi-step environments. The dominant narrative attributes agent performance to the underlying model. **This survey challenges that assumption.**
+
+We introduce a formal definition of the **agent execution harness** as a six-component tuple:
+
+| Component | Symbol | Role |
+|-----------|--------|------|
+| Execution Loop | **E** | Observe-think-act cycle, termination conditions, error recovery |
+| Tool Registry | **T** | Typed tool catalog, routing, monitoring, schema validation |
+| Context Manager | **C** | What enters the context window, compaction, retrieval |
+| State Store | **S** | Persistence across turns/sessions, crash recovery |
+| Lifecycle Hooks | **L** | Auth, logging, policy enforcement, instrumentation |
+| Evaluation Interface | **V** | Action trajectories, intermediate states, success signals |
+
+**Key empirical evidence that harnesses matter:**
+- 🔥 Pi Research: Grok Code Fast 1 jumped from **6.7% → 68.3%** on SWE-bench by changing *only* the harness edit-tool format — model unchanged
+- 💀 OpenAI Codex: **1M lines of code, 0 hand-written** over 5 months — failure attributed not to model capability but to "underspecified environments"
+- ⚡ Stripe Minions: **1,300 PRs/week, 0 human-written code** — harness-first engineering
+- 📉 METR: benchmark-passing PRs have a **24.2pp lower** human merge rate, gap widening at 9.6pp/year — evaluation harness validity crisis
+- 💬 *"The harness is the chassis; the model is the engine."* — practitioner consensus, 2026
+
+<p align="center">
+  <img src="assets/root_cause_diagram.png" width="640" alt="Root Cause Analysis"/>
+</p>
+
+### What This Survey Accomplishes
+
+**Conceptual contribution:** We formalize the agent harness as an architectural object with six governable components (E, T, C, S, L, V), elevating it from implicit infrastructure to an explicit research target.
+
+**Empirical scope:** We systematically review 110+ papers spanning academic research (evaluation benchmarks, security frameworks, memory architectures) and production deployments (Stripe, OpenAI, Cursor, METR), establishing that harness design is a binding constraint on deployed agent reliability.
+
+**Methodological advance:** We introduce the **Harness Completeness Matrix** — a structured assessment framework mapping which of the six components each system implements — enabling direct comparison across heterogeneous agent systems that prior surveys could not evaluate on common terms.
+
+**Open challenges identified:** We document nine technical challenges where current research provides partial solutions but no production-grade infrastructure: formal security models, cross-harness portability, protocol interoperability (MCP/A2A), context economics at 1M+ tokens/task, Byzantine fault tolerance in multi-agent systems, and compositional verification.
+
+**Practitioner-academic bridge:** Unlike prior surveys focused exclusively on model capabilities or isolated components (memory, planning, tool use), we synthesize peer-reviewed research with production deployment reports to show where theory meets practice — and where critical gaps remain.
+
+**Intended audience:** Researchers designing agent infrastructure, practitioners building production systems, and evaluators seeking to understand why benchmark performance often fails to predict deployment outcomes.
+
+---
+
+## Historical Timeline
+
+<p align="center">
+  <img src="assets/timeline.png" width="720" alt="Historical Evolution of Agent Harnesses"/>
+</p>
+
+| Year | Milestone | Significance |
+|------|-----------|--------------|
+| 1997–2005 | JUnit, TestNG, xUnit family | Software test harness paradigm; standardized observe-assert lifecycle |
+| 2016 | OpenAI Gym (Brockman et al.) | RL environment harness; step/reset API becomes canonical interface |
+| 2022 Nov | ChatGPT public release; LangChain emerges | LLM-native agent frameworks begin; tool-use as first-class citizen |
+| 2023 | ReAct, Toolformer, MemGPT, Reflexion, Voyager, AutoGPT | Core agent patterns: reasoning-acting, memory, reflection, skill accumulation |
+| 2023 | CAMEL, ChatDev, Generative Agents | Multi-agent coordination; social simulation harnesses |
+| 2023 | AgentBench, SWE-bench | Agent evaluation infrastructure emerges |
+| 2024 | MetaGPT, WebArena, ToolLLM, SWE-agent, OSWorld | Full-stack harnesses; real-world environment benchmarks |
+| 2024 | CodeAct, LATS, Tree of Thoughts | Structured action spaces; search-augmented planning |
+| 2024 Nov | Anthropic releases MCP protocol | First major tool↔harness standardization (2–15ms latency) |
+| 2025 | HAL, AIOS, LangGraph | Benchmark unification (21,730 rollouts); OS-level scheduling (2.1× speedup) |
+| 2025 | Google releases A2A protocol | Agent↔agent standardization (50–200ms) |
+| 2025 | MemoryOS, SkillsBench†, AgentBound† | Memory OS abstraction; skills-as-context (+16.2pp); safety certification |
+| 2026 Jan–Mar | AgencyBench†, SandboxEscapeBench†, PRISM†, AEGIS†, SkillFortify†, Schema First† | Compute economics; 15–35% escape rates; runtime security; schema discipline |
+
+*† preprint*
+
+---
+
+## Harness Completeness Matrix
+
+**Legend:** ✓ full support · ≈ partial · ✗ absent
+
+<table align="center">
+  <thead>
+    <tr>
+      <th>Category</th>
+      <th>System</th>
+      <th title="Execution Loop">E</th>
+      <th title="Tool Registry">T</th>
+      <th title="Context Manager">C</th>
+      <th title="State Store">S</th>
+      <th title="Lifecycle Hooks">L</th>
+      <th title="Evaluation Interface">V</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td rowspan="4"><strong>Full-Stack<br>Harnesses</strong></td>
+      <td>Claude Code</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>OpenClaw / PRISM</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td>
+    </tr>
+    <tr>
+      <td>AIOS</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>OpenHands</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>≈</td>
+    </tr>
+    <tr>
+      <td rowspan="6"><strong>Multi-Agent<br>Harnesses</strong></td>
+      <td>MetaGPT</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>AutoGen</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>ChatDev</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>CAMEL</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>DeerFlow</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>��</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>DeepAgents</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td rowspan="3"><strong>General<br>Frameworks</strong></td>
+      <td>LangChain</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>✗</td>
+    </tr>
+    <tr>
+      <td>LangGraph</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✗</td><td>✗</td>
+    </tr>
+    <tr>
+      <td>LlamaIndex</td>
+      <td>≈</td><td>✓</td><td>✓</td><td>≈</td><td>✗</td><td>✗</td>
+    </tr>
+    <tr>
+      <td rowspan="1"><strong>Specialized<br>Harnesses</strong></td>
+      <td>SWE-agent</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>✓</td>
+    </tr>
+    <tr>
+      <td rowspan="5"><strong>Capability<br>Modules</strong></td>
+      <td>MemGPT</td>
+      <td>✗</td><td>✗</td><td>✓</td><td>✓</td><td>✗</td><td>✗</td>
+    </tr>
+    <tr>
+      <td>Voyager</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>✓</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>Reflexion</td>
+      <td>≈</td><td>✗</td><td>≈</td><td>✓</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>Generative Agents</td>
+      <td>✓</td><td>✗</td><td>≈</td><td>✓</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>Concordia</td>
+      <td>✓</td><td>✗</td><td>≈</td><td>✓</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td rowspan="4"><strong>Evaluation<br>Infrastructure</strong></td>
+      <td>HAL</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✓</td>
+    </tr>
+    <tr>
+      <td>AgentBench</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✗</td><td>✓</td>
+    </tr>
+    <tr>
+      <td>OSWorld</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✗</td><td>✓</td>
+    </tr>
+    <tr>
+      <td>BrowserGym</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>✗</td><td>✓</td>
+    </tr>
+  </tbody>
+</table>
+
+---
+
+## Paper List
+
+### Historical Lineages
+
+#### Software Test Harnesses (1990s–2000s)
+
+- <u>JUnit</u>: **"JUnit: A Cook's Tour"**. *Beck & Gamma.* Java Report, 4(5), May 1999. [[Article](http://junit.sourceforge.net/doc/cookstour/cookstour.htm)]
+
+#### RL Environment Harnesses (2016–2022)
+
+- <u>OpenAI Gym</u>: **"OpenAI Gym"**. *Brockman et al.* arXiv 2016. [[Paper](https://arxiv.org/abs/1606.01540)] [[Code](https://github.com/openai/gym)]
+- <u>Gymnasium</u>: **"Gymnasium: A Standard Interface for Reinforcement Learning Environments"**. *Towers et al.* NeurIPS 2025. [[Paper](https://arxiv.org/abs/2407.17032)] [[Code](https://github.com/Farama-Foundation/Gymnasium)]
+
+#### Early LLM Agent Frameworks (2023–2024)
+
+- <u>ReAct</u>: **"ReAct: Synergizing Reasoning and Acting in Language Models"**. *Yao et al.* ICLR 2023. [[Paper](https://arxiv.org/abs/2210.03629)] [[Code](https://github.com/ysymyth/ReAct)]
+- <u>Toolformer</u>: **"Toolformer: Language Models Can Teach Themselves to Use Tools"**. *Schick et al.* NeurIPS 2023. [[Paper](https://arxiv.org/abs/2302.04761)]
+- <u>AutoGPT</u>: **"Auto-GPT: An Autonomous GPT-4 Experiment"**. *Gravitas et al.* GitHub 2023. [[Code](https://github.com/Significant-Gravitas/AutoGPT)]
+- <u>LangChain</u>: **"LangChain: Building Applications with LLMs through Composability"**. *Chase et al.* GitHub 2022. [[Code](https://github.com/langchain-ai/langchain)]
+
+---
+
+### Harness Taxonomy
+
+**What we classify:** We categorize agent systems by **harness completeness** — which of the six components (E, T, C, S, L, V) each system implements — distinguishing full-stack harnesses (all six components) from specialized frameworks (partial implementations).
+
+**Why it matters:** Prior taxonomies classified agents by application domain (coding, web navigation, embodied AI) or model architecture (single-agent, multi-agent). These categorizations cannot explain why systems with similar models achieve different reliability outcomes. Our harness-centric taxonomy reveals that production-grade systems converge on full ETCSLV implementations, while research prototypes often implement only 2-3 components.
+
+**Key finding:** No agent framework can achieve production reliability without implementing **all six governance components**. Systems missing L-component (lifecycle hooks) cannot enforce safety policies. Systems missing V-component (evaluation interfaces) cannot debug failures. Systems missing S-component (state persistence) cannot recover from crashes.
+
+#### Full-Stack Harnesses
+
+- <u>PRISM/OpenClaw</u>: **"OpenClaw PRISM: A Zero-Fork, Defense-in-Depth Runtime Security Layer for Tool-Augmented LLM Agents"**. *Li.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.11853)]
+- <u>AIOS</u>: **"AIOS: LLM Agent Operating System"**. *Mei et al.* COLM 2025. [[Paper](https://arxiv.org/abs/2403.16971)] [[Code](https://github.com/agiresearch/AIOS)]
+- <u>OpenHands</u>: **"OpenHands: An Open Platform for AI Software Developers as Generalist Agents"**. *Wang et al.* ICLR 2025. [[Paper](https://arxiv.org/abs/2407.16741)] [[Code](https://github.com/All-Hands-AI/OpenHands)]
+- <u>SWE-agent</u>: **"SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering"**. *Yang et al.* NeurIPS 2024. [[Paper](https://arxiv.org/abs/2405.15793)] [[Code](https://github.com/SWE-agent/SWE-agent)]
+- <u>HAL</u>: **"Holistic Agent Leaderboard: The Missing Infrastructure for AI Agent Evaluation"**. *Kapoor et al.* ICLR 2026. [[Paper](https://arxiv.org/abs/2510.11977)]
+
+#### Multi-Agent Harnesses
+
+- <u>MetaGPT</u>: **"MetaGPT: Meta Programming for a Multi-Agent Collaborative Framework"**. *Hong et al.* ICLR 2024. [[Paper](https://arxiv.org/abs/2308.00352)] [[Code](https://github.com/geekan/MetaGPT)]
+- <u>AutoGen</u>: **"AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation"**. *Wu et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2308.08155)] [[Code](https://github.com/microsoft/autogen)]
+- <u>ChatDev</u>: **"ChatDev: Communicative Agents for Software Development"**. *Qian et al.* ACL 2024. [[Paper](https://arxiv.org/abs/2307.07924)] [[Code](https://github.com/OpenBMB/ChatDev)]
+- <u>CAMEL</u>: **"CAMEL: Communicative Agents for 'Mind' Exploration of Large Language Model Society"**. *Li et al.* NeurIPS 2023. [[Paper](https://arxiv.org/abs/2303.17760)] [[Code](https://github.com/camel-ai/camel)]
+
+#### Frameworks & Modules
+
+- <u>LangGraph</u>: **"LangGraph: Build Resilient Language Agents as Graphs"**. *LangChain team.* GitHub 2024. [[Code](https://github.com/langchain-ai/langgraph)]
+- <u>MemGPT</u>: **"MemGPT: Towards LLMs as Operating Systems"**. *Packer et al.* NeurIPS 2023. [[Paper](https://arxiv.org/abs/2310.08560)] [[Code](https://github.com/cpacker/MemGPT)]
+- <u>Voyager</u>: **"Voyager: An Open-Ended Embodied Agent with Large Language Models"**. *Wang et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2305.16291)] [[Code](https://github.com/MineDojo/Voyager)]
+- <u>Reflexion</u>: **"Reflexion: Language Agents with Verbal Reinforcement Learning"**. *Shinn et al.* NeurIPS 2023. [[Paper](https://arxiv.org/abs/2303.11366)] [[Code](https://github.com/noahshinn/reflexion)]
+- <u>Generative Agents</u>: **"Generative Agents: Interactive Simulacra of Human Behavior"**. *Park et al.* UIST 2023. [[Paper](https://arxiv.org/abs/2304.03442)] [[Code](https://github.com/joonspk-research/generative_agents)]
+- <u>LangChain</u>: **"LangChain: Building Applications with LLMs through Composability"**. *Chase et al.* GitHub 2022. [[Code](https://github.com/langchain-ai/langchain)]
+- <u>LlamaIndex</u>: **"LlamaIndex: A Data Framework for LLM Applications"**. *Liu et al.* GitHub 2022. [[Code](https://github.com/run-llama/llama_index)]
+- <u>DeerFlow</u>: **"DeerFlow: Distributed Workflow Engine for LLM Agents"**. *GitHub 2024.* [[Code](https://github.com/modelscope/DeerFlow)]
+- <u>DeepAgents</u>: **"DeepAgents: Multi-Agent Framework for Deep Learning"**. *GitHub 2024.* [[Code](https://github.com/deepagents/deepagents)]
+
+#### Evaluation Infrastructure
+
+- <u>AgentBench</u>: **"AgentBench: Evaluating LLMs as Agents"**. *Liu et al.* ICLR 2024. [[Paper](https://arxiv.org/abs/2308.03688)] [[Code](https://github.com/THUDM/AgentBench)]
+- <u>SWE-bench</u>: **"SWE-bench: Can Language Models Resolve Real-World GitHub Issues?"**. *Jimenez et al.* ICLR 2024. [[Paper](https://arxiv.org/abs/2310.06770)] [[Code](https://github.com/swebench/SWE-bench)]
+- <u>OSWorld</u>: **"OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments"**. *Xie et al.* NeurIPS 2024. [[Paper](https://arxiv.org/abs/2404.07972)] [[Code](https://github.com/xlang-ai/OSWorld)]
+- <u>WebArena</u>: **"WebArena: A Realistic Web Environment for Building Autonomous Agents"**. *Zhou et al.* ICLR 2024. [[Paper](https://arxiv.org/abs/2307.13854)] [[Code](https://github.com/web-arena-x/webarena)]
+- <u>GAIA</u>: **"GAIA: A Benchmark for General AI Assistants"**. *Mialon et al.* ICLR 2024. [[Paper](https://arxiv.org/abs/2311.12983)]
+- <u>Mind2Web</u>: **"Mind2Web: Towards a Generalist Agent for the Web"**. *Deng et al.* NeurIPS 2023. [[Paper](https://arxiv.org/abs/2306.06070)]
+- <u>AgentBoard</u>: **"AgentBoard: An Analytical Evaluation Board of Multi-Turn LLM Agents"**. *Ma et al.* NeurIPS 2024. [[Paper](https://arxiv.org/abs/2401.13178)]
+- <u>InterCode</u>: **"InterCode: Standardizing and Benchmarking Interactive Coding with Execution Feedback"**. *Yang et al.* NeurIPS 2023. [[Paper](https://arxiv.org/abs/2306.14898)]
+
+---
+
+### Technical Challenges
+
+**Scope:** We identify nine open research problems where current harness infrastructure provides partial solutions but no production-grade, generalizable frameworks. Each challenge represents a gap between what academic prototypes demonstrate and what deployed systems require.
+
+**Evaluation criteria:** For each challenge, we assess: (1) **Academic coverage** — peer-reviewed solutions, if any; (2) **Production maturity** — whether any deployed system has solved it at scale; (3) **Blocking severity** — whether the unsolved problem prevents deployment or merely limits performance.
+
+**Cross-cutting theme:** Most challenges stem from the **composition problem** — individual components (security sandboxes, memory stores, evaluation metrics) work in isolation, but combining them into a coherent harness surfaces emergent failures. Formal verification, portability testing, and protocol interoperability all require compositional reasoning that current research has not addressed.
+
+#### Security & Sandboxing
+
+**Challenge:** Agent sandboxes must prevent malicious tool use while permitting legitimate system access — a harder problem than traditional application sandboxing because agents **intentionally** interact with sensitive resources (file systems, network APIs, shell commands).
+
+**State of the art:** SandboxEscapeBench shows frontier models can escape containers; PRISM demonstrates zero-fork defense-in-depth is possible but requires OS-level instrumentation. No generalizable sandbox exists that works across all agent use cases without manual per-task configuration.
+
+**Open problems:** Compositional security properties (proving agent + harness + environment satisfy a safety specification) remain unsolved. Current defenses are reactive (detecting escapes) rather than preventive (proving impossibility of escape).
+
+> **Key numbers:** SandboxEscapeBench — frontier LLMs achieve **15–35% container escape rates**; PRISM — 10-hook zero-fork runtime reduces escape to near-zero with <5ms overhead.
+
+- <u>SandboxEscapeBench</u>†: **"Quantifying Frontier LLM Capabilities for Container Sandbox Escape"**. *Marchand et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.02277)]
+- <u>InjecAgent</u>: **"InjecAgent: Benchmarking Indirect Prompt Injections in Tool-Integrated Large Language Model Agents"**. *Zhan et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2403.02691)]
+- <u>ToolHijacker</u>†: **"Prompt Injection Attack to Tool Selection in LLM Agents"**. *Shi et al.* NDSS 2026. [[Paper](https://arxiv.org/abs/2504.19793)]
+- <u>Securing MCP</u>†: **"Securing the Model Context Protocol (MCP): Risks, Controls, and Governance"**. *Errico et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2511.20920)]
+- <u>SHIELDA</u>†: **"SHIELDA: Structured Handling of Exceptions in LLM-Driven Agentic Workflows"**. *Zhou et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2508.07935)]
+- <u>PALADIN</u>†: **"PALADIN: Self-Correcting Language Model Agents to Cure Tool-Failure Cases"**. *Vuddanti et al.* ICLR 2026. [[Paper](https://arxiv.org/abs/2509.25238)]
+- <u>AgentBound</u>†: **"Securing AI Agent Execution"**. *Bühler et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2510.21236)]
+- <u>AgentSys</u>†: **"AgentSys: Secure and Dynamic LLM Agents Through Explicit Hierarchical Memory Management"**. *Wen et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2602.07398)]
+- <u>Indirect Prompt Injection</u>: **"Not What You've Signed Up For: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection"**. *Greshake et al.* AISec 2023. [[Paper](https://arxiv.org/abs/2302.12173)]
+- <u>AgentHarm</u>†: **"AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents"**. *Andriushchenko et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2410.09024)]
+- <u>TrustAgent</u>: **"TrustAgent: Towards Safe and Trustworthy LLM-Based Agents"**. *Hua et al.* EMNLP 2024. [[Paper](https://arxiv.org/abs/2402.01586)]
+- <u>ToolEmu</u>†: **"Identifying the Risks of LM Agents with an LM-Emulated Sandbox"**. *Ruan et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2309.15817)]
+- <u>Ignore Previous Prompt</u>: **"Ignore Previous Prompt: Attack Techniques For Language Models"**. *Perez & Ribeiro.* NeurIPS ML Safety Workshop 2022. [[Paper](https://arxiv.org/abs/2211.09527)]
+
+#### Evaluation & Benchmarking
+
+> **Key numbers:** HAL unified **21,730 rollouts**, compressing weeks of evaluation to hours; OSWorld reports **28% false negative rate** in automated evaluation; METR finds benchmark-passing PRs have **24.2pp lower** human merge rate, widening at 9.6pp/year.
+
+- <u>AgencyBench</u>†: **"AgencyBench: Benchmarking the Frontiers of Autonomous Agents in 1M-Token Real-World Contexts"**. *Li et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2601.11044)]
+- <u>AEGIS</u>†: **"AEGIS: No Tool Call Left Unchecked -- A Pre-Execution Firewall and Audit Layer for AI Agents"**. *Yuan et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.12621)]
+- <u>Hell or High Water</u>†: **"Hell or High Water: Evaluating Agentic Recovery from External Failures"**. *Wang et al.* COLM 2025. [[Paper](https://arxiv.org/abs/2508.11027)]
+- <u>SearchLLM</u>†: **"Aligning Large Language Models with Searcher Preferences"**. *Wu et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.10473)]
+- <u>Meta-Harness</u>†: **"Meta-Harness: End-to-End Optimization of Model Harnesses"**. *Lee et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.28052)]
+- <u>TheAgentCompany</u>†: **"TheAgentCompany: Benchmarking LLM Agents on Consequential Real-World Tasks"**. *Xu et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2412.14161)]
+- <u>BrowserGym</u>†: **"The BrowserGym Ecosystem for Web Agent Research"**. *Le Sellier De Chezelles et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2412.05467)]
+- <u>WorkArena</u>†: **"WorkArena: How Capable are Web Agents at Solving Common Knowledge Work Tasks?"**. *Drouin et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2403.07718)]
+- <u>R-Judge</u>: **"R-Judge: Benchmarking Safety Risk Awareness for LLM Agents"**. *Yuan et al.* EMNLP 2024. [[Paper](https://arxiv.org/abs/2401.10019)]
+- <u>R2E</u>: **"R2E: Turning any GitHub Repository into a Programming Agent Environment"**. *Jain et al.* ICML 2024. [[Paper](https://proceedings.mlr.press/v235/jain24c.html)]
+- <u>Evaluation Survey</u>: **"Evaluation and Benchmarking of LLM Agents: A Survey"**. *Mohammadi et al.* KDD 2025. [[Paper](https://arxiv.org/abs/2507.21504)]
+- <u>PentestJudge</u>†: **"PentestJudge: Judging Agent Behavior Against Operational Requirements"**. *Caldwell et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2508.02921)]
+
+#### Protocol Standardization
+
+> **Key numbers:** MCP (tool↔harness): 2–15ms latency; A2A (agent↔agent): 50–200ms; ACP (intent-level, IBM) — three protocols serve complementary roles.
+
+- <u>MCP</u>: **"Model Context Protocol"**. *Anthropic.* Technical Report 2024. [[Spec](https://modelcontextprotocol.io)]
+- <u>A2A</u>: **"Agent-to-Agent Protocol"**. *Google.* Technical Report 2025. [[Spec](https://google.github.io/A2A/)]
+- <u>Protocol Comparison</u>†: **"A Survey of Agent Interoperability Protocols: Model Context Protocol (MCP), Agent Communication Protocol (ACP), Agent-to-Agent Protocol (A2A), and Agent Network Protocol (ANP)"**. *Ehtesham et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2505.02279)]
+- <u>Gorilla</u>: **"Gorilla: Large Language Model Connected with Massive APIs"**. *Patil et al.* NeurIPS 2023. [[Paper](https://arxiv.org/abs/2305.15334)] [[Code](https://github.com/ShishirPatil/gorilla)]
+
+#### Runtime Context Management
+
+> **Key numbers:** SkillsBench — curated skill injection yields **+16.2pp** improvement; "Lost in the Middle" effect documented; long-context models shift the problem from *retention* to *salience*.
+
+- <u>SkillsBench</u>†: **"SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks"**. *Li et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2602.12670)]
+- <u>ReadAgent</u>: **"A Human-Inspired Reading Agent with Gist Memory of Very Long Contexts"**. *Lee et al.* ICML 2024. [[Paper](https://arxiv.org/abs/2402.09727)]
+- <u>MemoryOS</u>: **"Memory OS of AI Agent"**. *Kang et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2506.06326)]
+- <u>CoALA</u>: **"Cognitive Architectures for Language Agents"**. *Sumers et al.* TMLR 2024. [[Paper](https://arxiv.org/abs/2309.02427)]
+- <u>SkillFortify</u>†: **"Formal Analysis and Supply Chain Security for Agentic AI Skills"**. *Bhardwaj.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.00195)]
+- <u>Lost in the Middle</u>: **"Lost in the Middle: How Language Models Use Long Contexts"**. *Liu et al.* TACL 2024. [[Paper](https://arxiv.org/abs/2307.03172)]
+- <u>Context Engineering Survey</u>†: **"Context Engineering: A Survey of 1,400 Papers on Effective Context Management for LLM Agents"**. *Mei et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2507.13334)]
+
+#### Tool Use & Registry
+
+> **Key numbers:** Vercel found removing **80% of tools** helped more than any model upgrade; Schema First (Sigdel & Baral, 2026) — a controlled pilot showing that schema-based tool contracts reduce *interface* misuse but not *semantic* misuse, with end-task success at zero across all conditions, suggesting interface design alone is insufficient for tool reliability; CodeAct outperforms on **17/17 Mint benchmarks** with **−20% turns**.
+
+- <u>CodeAct</u>: **"Executable Code Actions Elicit Better LLM Agents"**. *Wang et al.* ICML 2024. [[Paper](https://arxiv.org/abs/2402.01030)] [[Code](https://github.com/xingyaoww/code-act)]
+- <u>Schema First</u>†: **"Schema First Tool APIs for LLM Agents: A Controlled Study of Tool Misuse, Recovery, and Budgeted Performance"**. *Sigdel & Baral.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.13404)]
+- <u>ToolLLM</u>: **"ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs"**. *Qin et al.* ICLR 2024. [[Paper](https://arxiv.org/abs/2307.16789)] [[Code](https://github.com/OpenBMB/ToolBench)]
+- <u>ToolSandbox</u>†: **"ToolSandbox: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities"**. *Lu et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2408.04682)]
+- <u>AutoTool</u>†: **"AutoTool: Efficient Tool Selection for Large Language Model Agents"**. *Jia & Li.* AAAI 2026. [[Paper](https://arxiv.org/abs/2511.14650)]
+- <u>Tool Learning Survey</u>: **"Tool Learning with Large Language Models: A Survey"**. *Qu et al.* Frontiers of Computer Science 2024. [[Paper](https://arxiv.org/abs/2405.17935)]
+- <u>GoEX</u>†: **"GoEX: Perspectives and Designs Towards a Runtime for Autonomous LLM Applications"**. *Patil et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2404.06921)]
+- <u>AgentTuning</u>: **"AgentTuning: Enabling Generalized Agent Abilities for LLMs"**. *Zeng et al.* ACL 2024. [[Paper](https://arxiv.org/abs/2310.12823)]
+
+#### Memory Architecture
+
+> **Key numbers:** Mem0 achieves **90% token reduction** vs full-context; Zep temporal knowledge: **+18.5% QA accuracy**; Agent Workflow Memory: **+14.9%** on Mind2Web. Six architectural patterns: flat buffer → hierarchical → episodic → semantic → procedural → graph.
+
+- <u>Agent Workflow Memory (AWM)</u>†: **"Agent Workflow Memory"**. *Wang et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2409.07429)]
+- <u>Mem0</u>†: **"Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory"**. *Khant et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2504.19413)]
+- <u>A-MEM</u>†: **"A-MEM: Agentic Memory for LLM Agents"**. *Xu et al.* NeurIPS 2025. [[Paper](https://arxiv.org/abs/2502.12110)]
+- <u>MemAct</u>†: **"Memory as Action: Autonomous Context Curation for Long-Horizon Agentic Tasks"**. *Zhang et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2510.12635)]
+- <u>Memory Survey</u>†: **"Memory for Autonomous LLM Agents: Mechanisms, Evaluation, and Emerging Frontiers"**. *Du.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.07670)]
+- <u>MemoryBank</u>: **"MemoryBank: Enhancing Large Language Models with Long-Term Memory"**. *Zhong et al.* AAAI 2024. [[Paper](https://arxiv.org/abs/2305.10250)]
+- <u>LoCoMo</u>†: **"Evaluating Very Long-Term Conversational Memory of LLM Agents"**. *Maharana et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2402.17753)]
+- <u>Memory Mechanisms Survey</u>†: **"A Survey on the Memory Mechanism of Large Language Model Based Agents"**. *Zhang et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2404.13501)]
+- <u>Evo-Memory</u>†: **"Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory"**. *Wei et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2511.20857)]
+
+#### Planning & Reasoning
+
+> **Key numbers:** SWE-agent ACI study shows interface design outweighs model capability as the primary performance determinant. LATS integrates MCTS with language model feedback for state-space search.
+
+- <u>Tree of Thoughts</u>: **"Tree of Thoughts: Deliberate Problem Solving with Large Language Models"**. *Yao et al.* NeurIPS 2023. [[Paper](https://arxiv.org/abs/2305.10601)] [[Code](https://github.com/princeton-nlp/tree-of-thought-llm)]
+- <u>LATS</u>: **"Language Agent Tree Search Unifies Reasoning, Acting, and Planning in Language Models"**. *Zhou et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2310.04406)] [[Code](https://github.com/lapisrocks/LanguageAgentTreeSearch)]
+- <u>AFlow</u>†: **"AFlow: Automating Agentic Workflow Generation"**. *Zhang et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2410.10762)]
+- <u>Agent Q</u>†: **"Agent Q: Advanced Reasoning and Learning for Autonomous AI Agents"**. *Putta et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2408.07199)]
+- <u>OPENDEV</u>†: **"Building Effective AI Coding Agents for the Terminal: Scaffolding, Harness, Context Engineering, and Lessons Learned"**. *Bui.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.05344)]
+- <u>AOrchestra</u>†: **"AOrchestra: Automating Sub-Agent Creation for Agentic Orchestration"**. *Ruan et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2602.03786)]
+- <u>RAP</u>: **"Reasoning with Language Model is Planning with World Model"**. *Hao et al.* EMNLP 2023. [[Paper](https://arxiv.org/abs/2305.14992)]
+- <u>Inner Monologue</u>: **"Inner Monologue: Embodied Reasoning Through Planning with Language Models"**. *Huang et al.* CoRL 2022. [[Paper](https://arxiv.org/abs/2207.05608)]
+- <u>Agent-Oriented Planning</u>: **"Agent-Oriented Planning in Multi-Agent Systems"**. *Li et al.* ICLR 2025. [[Paper](https://arxiv.org/abs/2410.02189)]
+- <u>ExACT</u>†: **"ExACT: Teaching AI Agents to Explore with Reflective-MCTS and Exploratory Learning"**. *Yu et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2410.02052)]
+
+#### Multi-Agent Coordination
+
+> **Key numbers:** AgencyBench — agents achieve **48.4% success on native SDK harness** vs substantially lower on independent harnesses, demonstrating tight harness-agent coupling. Byzantine fault tolerance remains an open problem for adversarial multi-agent settings.
+
+<p align="center">
+  <img src="assets/multi_agent_topology.png" width="680" alt="Multi-Agent Coordination Topologies"/>
+</p>
+
+- <u>SAGA</u>†: **"SAGA: A Security Architecture for Governing AI Agentic Systems"**. *Syros et al.* NDSS 2026. [[Paper](https://arxiv.org/abs/2504.21034)]
+- <u>MAS-FIRE</u>†: **"MAS-FIRE: Fault Injection and Reliability Evaluation for LLM-Based Multi-Agent Systems"**. *Jia et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2602.19843)]
+- <u>Byzantine fault tolerance</u>†: **"Rethinking the Reliability of Multi-agent System: A Perspective from Byzantine Fault Tolerance"**. *Zheng et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2511.10400)]
+- <u>Multi-agent baseline study</u>†: **"Rethinking the Value of Multi-Agent Workflow: A Strong Single Agent Baseline"**. *Xu et al.* arXiv 2026. [[Paper](https://arxiv.org/abs/2601.12307)]
+- <u>AgentVerse</u>†: **"AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors"**. *Chen et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2308.10848)]
+- <u>Mixture-of-Agents</u>†: **"Mixture-of-Agents Enhances Large Language Model Capabilities"**. *Wang et al.* arXiv 2024. [[Paper](https://arxiv.org/abs/2406.04692)]
+- <u>Multi-Agent Survey</u>: **"Large Language Model Based Multi-Agents: A Survey of Progress and Challenges"**. *Guo et al.* IJCAI 2024. [[Paper](https://arxiv.org/abs/2402.01680)]
+- <u>Concordia</u>†: **"Generative Agent-Based Modeling with Actions Grounded in Physical, Social, or Digital Space Using Concordia"**. *Vezhnevets et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2312.03664)]
+
+#### Compute Economics
+
+> **Key numbers:** OpenRouter reports **13T tokens/week** (Feb 2026), doubling every 4 weeks; AgencyBench measures **1M tokens/task** average; 1000× agent compute growth projected by 2027; AIOS achieves **2.1× throughput speedup** via proper agent scheduling.
+
+- <u>Repo2Run</u>†: **"Repo2Run: Automated Building Executable Environment for Code Repository at Scale"**. *Hu et al.* arXiv 2025. [[Paper](https://arxiv.org/abs/2502.13681)]
+- <u>Policy-First</u>†: **"Guardrails as Infrastructure: Policy-First Control for Tool-Orchestrated Workflows"**. *Sigdel & Baral.* arXiv 2026. [[Paper](https://arxiv.org/abs/2603.18059)]
+
+---
+
+### Emerging Topics
+
+- <u>Self-Evolving Agents Survey</u>†: **"A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence"**. *Gao et al.* TMLR 2026. [[Paper](https://arxiv.org/abs/2507.21046)]
+- <u>Self-RAG</u>: **"Self-RAG: Learning to Retrieve, Generate, and Critique Through Self-Reflection"**. *Asai et al.* ICLR 2024. [[Paper](https://arxiv.org/abs/2310.11511)]
+- <u>Constitutional AI</u>: **"Constitutional AI: Harmlessness from AI Feedback"**. *Bai et al.* arXiv 2022. [[Paper](https://arxiv.org/abs/2212.08073)]
+- <u>AppAgent</u>†: **"AppAgent: Multimodal Agents as Smartphone Users"**. *Zhang et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2312.13771)]
+
+#### Related Surveys
+
+- <u>LLM Agents Survey</u>: **"A Survey on Large Language Model Based Autonomous Agents"**. *Wang et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2308.11432)]
+- <u>Rise of LLM Agents</u>: **"The Rise and Potential of Large Language Model Based Agents: A Survey"**. *Xi et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2309.07864)]
+- <u>LLM Survey</u>: **"A Survey of Large Language Models"**. *Zhao et al.* arXiv 2023. [[Paper](https://arxiv.org/abs/2303.18223)]
+- <u>AI Agent Systems</u>†: **"AI Agent Systems: Architectures, Applications, and Evaluation"**. *Xu.* arXiv 2025. [[Paper](https://arxiv.org/abs/2601.01743)]
+
+#### Practitioner Reports & Industry Insights
+
+> Production deployment experiences from Stripe, OpenAI, Cursor, METR, and other frontier practitioners.
+
+- <u>Stripe Minions</u>: **"Minions: Stripe's one-shot, end-to-end coding agents"**. *Gray.* Stripe Dev Blog, Feb 2026. [[Blog](https://stripe.dev/blog/minions-stripes-one-shot-end-to-end-coding-agents)]
+- <u>Harness Engineering (OpenAI)</u>: **"Harness engineering: leveraging Codex in an agent-first world"**. *Lopopolo.* OpenAI Blog, Feb 2026. [[Blog](https://openai.com/index/harness-engineering/)]
+- <u>Self-Driving Codebases</u>: **"Towards self-driving codebases"**. *Lin.* Cursor Blog, Feb 2026. [[Blog](https://cursor.com/blog/self-driving-codebases)]
+- <u>METR SWE-bench Analysis</u>: **"Many SWE-bench-Passing PRs Would Not Be Merged into Main"**. *Whitfill et al.* METR Notes, Mar 2026. [[Report](https://metr.org/notes/2026-03-10-many-swe-bench-passing-prs-would-not-be-merged-into-main/)]
+
+---
+
+### Future Directions
+
+Eight open research directions identified in the survey (no curated paper list — these are forward-looking):
+
+- **Formal Harness Specification Language** — DSL for specifying and verifying H=(E,T,C,S,L,V) components
+- **Cross-Harness Benchmark Suite** — portability testing across incompatible harness ecosystems
+- **Security Taxonomy & Threat Model** — extension of OWASP Top 10 to agent harness attack surfaces
+- **Protocol Interoperability (MCP/A2A)** — bridging tool-level and agent-level protocols
+- **Long-Horizon Evaluation Methodology** — metrics that don't degrade under multi-session, multi-day tasks
+- **Harness-Aware Fine-Tuning** — training models that are aware of their execution environment
+- **Memory Interface Standardization** — portable memory APIs across flat, episodic, and graph stores
+- **Harness Transparency Specification** — auditability and explainability for runtime decisions
+
+---
+
+## Citation
+
+See BibTeX at the top of this README.
+
+---
+
+## Update Log
+
+| Version | Date | Changes |
+|---------|------|---------|
+|  v1 | 2026-04-03 | Initial preprint |
+|  v2 | 2026-04-07 | Repo updated |
+
+---
+
+<p align="center">
+  <i>† denotes preprint, not yet peer-reviewed.</i><br>
+  <i>This survey is under active development; the full manuscript will be released soon.</i><br>
+  <i>Maintained by Qianyu Meng & Liyi Chen. PRs welcome for missing papers or updated links.</i>
+</p>

README_zh.md

@@ -0,0 +1,474 @@
+---
+license: cc-by-4.0
+language:
+  - en
+  - zh
+tags:
+  - survey
+  - llm-agents
+  - agent-harness
+  - agent-frameworks
+  - harness-engineering
+  - evaluation
+  - agent-memory
+  - multi-agent-systems
+pretty_name: "Agent Harness for Large Language Model Agents: A Survey"
+size_categories:
+  - n<1K
+---
+
+[English](README.md) | [中文](README_zh.md)
+
+# 大语言模型智能体约束工程框架综述
+
+<p align="center">
+  <a href="https://github.com/Gloriaameng/LLM-Agent-Harness-Survey/stargazers"><img src="https://img.shields.io/github/stars/Gloriaameng/LLM-Agent-Harness-Survey?style=social" alt="GitHub Stars"></a>
+  <a href="LICENSE"><img src="https://img.shields.io/badge/License-CC--BY--4.0-blue.svg" alt="License"></a>
+  <img src="https://img.shields.io/badge/Papers-110%2B-green" alt="Papers">
+  <img src="https://img.shields.io/badge/Version-v2-orange" alt="Version">
+</p>
+
+<p align="center">
+  <img src="assets/architecture_diagram.png" width="720" alt="H=(E,T,C,S,L,V) 六组件架构"/>
+</p>
+
+> ⭐ **本 survey 持续更新中。如果对你有帮助，欢迎 Star 关注最新进展，也帮助更多人发现它。**
+
+---
+
+> **智能体约束工程框架——而非模型本身——是智能体大规模部署可靠性的首要决定因素。**  
+> 本综述将约束工程框架形式化为一级架构对象 **H = (E, T, C, S, L, V)**，系统调研了涵盖23个系统的110余篇论文、博客和报告，并总结了9项开放性技术挑战。  
+> 📄 **[阅读论文](#)** （即将发布）  
+> ✉️ 勘误与建议：gloriamenng@gmail.com; chenliyi@xiaohongshu.com
+
+如果本综述对您有所帮助，请引用：
+
+```bibtex
+@misc{meng2026agentharness,
+  title   = {Agent Harness for Large Language Model Agents: A Survey},
+  author  = {Meng, Qianyu and Wang, Yanan and Chen, Liyi and Wang, Qimeng and
+             Lu, Chengqiang and Wu, Wei and Gao, Yan and Wu, Yi and Hu, Yao},
+  year    = {2026},
+  url     = {https://github.com/Gloriaameng/LLM-Agent-Harness-Survey},
+  note    = {Work in progress}
+}
+```
+
+---
+
+## 🆕 新闻与更新
+
+- **[2026-04-03]** 首次发布
+- **[2026-04-07]** 仓库更新
+
+---
+
+## 目录
+
+- [概述](#概述)
+- [历史时间线](#历史时间线)
+- [约束工程框架完备性矩阵](#执行器完备性矩阵)
+- [论文列表](#论文列表)
+  - [历史渊源](#历史渊源)
+  - [约束工程框架分类](#执行器分类)
+  - [技术挑战](#技术挑战)
+    - [安全与沙箱](#安全与沙箱)
+    - [评估与基准测试](#评估与基准测试)
+    - [协议标准化](#协议标准化)
+    - [运行时上下文管理](#运行时上下文管理)
+    - [工具使用与注册](#工具使用与注册)
+    - [记忆架构](#记忆架构)
+    - [规划与推理](#规划与推理)
+    - [多智能体协同](#多智能体协同)
+    - [计算经济学](#计算经济学)
+  - [新兴主题](#新兴主题)
+  - [未来方向](#未来方向)
+- [引用](#引用)
+- [更新日志](#更新日志)
+
+---
+
+## 概述
+
+大语言模型（LLM）智能体正日益部署于自主规划、工具使用和多步骤交互环境中。主流叙事将智能体性能归因于底层模型能力。**本综述挑战了这一假设。**
+
+我们提出了**智能体约束工程框架**的形式化定义，将其定义为包含六个组件的元组：
+
+| 组件 | 符号 | 作用 |
+|------|------|------|
+| 执行循环 | **E** | 观察-思考-行动循环，终止条件，错误恢复 |
+| 工具注册表 | **T** | 类型化工具目录，路由，监控，模式校验 |
+| 上下文管理器 | **C** | 上下文窗口内容选择，压缩，检索 |
+| 状态存储 | **S** | 跨回合/会话持久化，崩溃恢复 |
+| 生命周期钩子 | **L** | 认证，日志，策略执行，监测埋点 |
+| 评估接口 | **V** | 行动轨迹，中间状态，成功信号 |
+
+**约束工程重要性的关键实证证据：**
+- 🔥 Pi Research：Grok Code Fast 1 模型在 SWE-bench 上从 **6.7% → 68.3%**，仅通过修改约束工程的编辑工具格式——模型不变
+- 💀 OpenAI Codex：5个月内生成**100万行代码，0行手写**——失败归因于"环境规范不足"而非模型能力
+- ⚡ Stripe Minions：**每周1300个PR，0行人工代码**——约束优先工程
+- 📉 METR：通过基准测试的PR人工合并率**低24.2个百分点**，差距以9.6pp/年速度扩大——评估约束工程框架有效性危机
+- 💬 *"约束工程是底盘；模型是引擎。"* ——2026年业界共识
+
+<p align="center">
+  <img src="assets/root_cause_diagram.png" width="640" alt="根因分析"/>
+</p>
+
+### 本综述的学术贡献
+
+**概念贡献：**我们将智能体约束工程框架形式化为具有六个可治理组件（E, T, C, S, L, V）的架构对象，将其从隐式基础设施提升为显式研究目标。
+
+**实证范围：**我们系统性地综述了110余篇论文，涵盖学术研究（评估基准、安全框架、记忆架构）和生产部署（Stripe、OpenAI、Cursor、METR）��确立了约束工程框架设计是已部署智能体可靠性的约束瓶颈这一观点。
+
+**方法论进展：**我们引入了**约束工程框架完备性矩阵**——一个结构化评估框架，映射每个系统实现了六个组件中的哪些——使得先前综述无法实现的异构智能体系统直接对比成为可能。
+
+**已识别的开放挑战：**我们记录了九项技术挑战，这些挑战当前研究仅提供了部分解决方案而无生产级基础设施：形式化安全模型、跨工程框架可移植性、协议互操作性（MCP/A2A）、百万级token/任务的上下文经济性、多智能体系统中的拜占庭容错、组合式验证。
+
+**学术界与业界的桥梁：**与先前仅关注模型能力或孤立组件（记忆、规划、工具使用）的综述不同，我们综合了同行评审研究与生产部署报告，展示了理论与实践的交汇点——以及仍存在的关键差距。
+
+**目标读者：**设计智能体基础设施的研究者、构建生产系统的实践者、以及试图理解为何基准测试性能常无法预测部署结果的评估者。
+
+---
+
+## 历史时间线
+
+<p align="center">
+  <img src="assets/timeline.png" width="720" alt="智能体约束工程的历史演进"/>
+</p>
+
+| 年份 | 里程碑 | 意义 |
+|------|--------|------|
+| 1997–2005 | JUnit, TestNG, xUnit 家族 | 软件测试约束工程范式；标准化观察-断言生命周期 |
+| 2016 | OpenAI Gym (Brockman et al.) | 强化学习环境约束工程框架；step/reset API 成为规范接口 |
+| 2022年11月 | ChatGPT 公开发布；LangChain 出现 | 面向LLM的智能体框架萌芽；工具使用成为一级公民 |
+| 2023 | ReAct, Toolformer, MemGPT, Reflexion, Voyager, AutoGPT | 核心智能体模式：推理-行动、记忆、反思、技能积累 |
+| 2023 | CAMEL, ChatDev, Generative Agents | 多智能体协同；社会仿真约束工程 |
+| 2023 | AgentBench, SWE-bench | 智能体评估基础设施涌现 |
+| 2024 | MetaGPT, WebArena, ToolLLM, SWE-agent, OSWorld | 全栈执行器；真实世界环境基准测试 |
+| 2024 | CodeAct, LATS, Tree of Thoughts | 结构化动作空间；搜索增强规划 |
+| 2024年11月 | Anthropic 发布 MCP 协议 | 首个主要工具↔约束工程标准化（2–15ms延迟） |
+| 2025 | HAL, AIOS, LangGraph | 基准测试统一（21,730次评估）；操作系统级调度（2.1×加速） |
+| 2025 | Google 发布 A2A 协议 | 智能体↔智能体标准化（50–200ms） |
+| 2025 | MemoryOS, SkillsBench†, AgentBound† | 记忆操作系统抽象；技能即上下文（+16.2pp）；安全认证 |
+| 2026年1–3月 | AgencyBench†, SandboxEscapeBench†, PRISM†, AEGIS†, SkillFortify†, Schema First† | 计算经济学；15–35%逃逸率；运行时安全；模式规范 |
+
+*† 表示预印本，尚未经同行评审*
+
+---
+
+## 约束工程框架完备性矩阵
+
+**图例：** ✓ 完全支持 · ≈ 部分支持 · ✗ 缺失
+
+<table align="center">
+  <thead>
+    <tr>
+      <th>类别</th>
+      <th>系统</th>
+      <th title="执行循环">E</th>
+      <th title="工具注册表">T</th>
+      <th title="上下文管理器">C</th>
+      <th title="状态存储">S</th>
+      <th title="生命周期钩子">L</th>
+      <th title="评估接口">V</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td rowspan="4"><strong>全栈式<br>约束工程</strong></td>
+      <td>Claude Code</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>OpenClaw / PRISM</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td>
+    </tr>
+    <tr>
+      <td>AIOS</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>OpenHands</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>✓</td><td>≈</td>
+    </tr>
+    <tr>
+      <td rowspan="6"><strong>多智能体<br>执行器</strong></td>
+      <td>MetaGPT</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>AutoGen</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>ChatDev</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>CAMEL</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>DeerFlow</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>DeepAgents</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>≈</td>
+    </tr>
+    <tr>
+      <td rowspan="3"><strong>通用<br>框架</strong></td>
+      <td>LangChain</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>✗</td>
+    </tr>
+    <tr>
+      <td>LangGraph</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✗</td><td>✗</td>
+    </tr>
+    <tr>
+      <td>LlamaIndex</td>
+      <td>≈</td><td>✓</td><td>✓</td><td>≈</td><td>✗</td><td>✗</td>
+    </tr>
+    <tr>
+      <td rowspan="1"><strong>专用<br>约束工程</strong></td>
+      <td>SWE-agent</td>
+      <td>✓</td><td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>✓</td>
+    </tr>
+    <tr>
+      <td rowspan="5"><strong>能力<br>模块</strong></td>
+      <td>MemGPT</td>
+      <td>✗</td><td>✗</td><td>✓</td><td>✓</td><td>✗</td><td>✗</td>
+    </tr>
+    <tr>
+      <td>Voyager</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>✓</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>Reflexion</td>
+      <td>≈</td><td>✗</td><td>≈</td><td>✓</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>Generative Agents</td>
+      <td>✓</td><td>✗</td><td>≈</td><td>✓</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td>Concordia</td>
+      <td>✓</td><td>✗</td><td>≈</td><td>✓</td><td>✗</td><td>≈</td>
+    </tr>
+    <tr>
+      <td rowspan="4"><strong>评估<br>基础设施</strong></td>
+      <td>HAL</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✓</td>
+    </tr>
+    <tr>
+      <td>AgentBench</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✗</td><td>✓</td>
+    </tr>
+    <tr>
+      <td>OSWorld</td>
+      <td>✓</td><td>≈</td><td>≈</td><td>≈</td><td>✗</td><td>✓</td>
+    </tr>
+    <tr>
+      <td>BrowserGym</td>
+      <td>✓</td><td>✓</td><td>≈</td><td>≈</td><td>✗</td><td>✓</td>
+    </tr>
+  </tbody>
+</table>
+
+---
+
+## 论文列表
+
+### 历史渊源
+
+#### 软件测试约束工程 (1990s–2000s)
+
+- <u>JUnit</u>: **"JUnit: A Cook's Tour"**. *Beck & Gamma.* Java Report, 4(5), May 1999. [[文章](http://junit.sourceforge.net/doc/cookstour/cookstour.htm)]
+
+#### 强化学习环境约束工程框架 (2016–2022)
+
+- <u>OpenAI Gym</u>: **"OpenAI Gym"**. *Brockman et al.* arXiv 2016. [[论文](https://arxiv.org/abs/1606.01540)] [[代码](https://github.com/openai/gym)]
+- <u>Gymnasium</u>: **"Gymnasium: A Standard Interface for Reinforcement Learning Environments"**. *Towers et al.* NeurIPS 2025. [[论文](https://arxiv.org/abs/2407.17032)] [[代码](https://github.com/Farama-Foundation/Gymnasium)]
+
+#### 早期LLM智能体框架 (2023–2024)
+
+- <u>ReAct</u>: **"ReAct: Synergizing Reasoning and Acting in Language Models"**. *Yao et al.* ICLR 2023. [[论文](https://arxiv.org/abs/2210.03629)] [[代码](https://github.com/ysymyth/ReAct)]
+- <u>Toolformer</u>: **"Toolformer: Language Models Can Teach Themselves to Use Tools"**. *Schick et al.* NeurIPS 2023. [[论文](https://arxiv.org/abs/2302.04761)]
+- <u>AutoGPT</u>: **"Auto-GPT: An Autonomous GPT-4 Experiment"**. *Gravitas et al.* GitHub 2023. [[代码](https://github.com/Significant-Gravitas/AutoGPT)]
+- <u>LangChain</u>: **"LangChain: Building Applications with LLMs through Composability"**. *Chase et al.* GitHub 2022. [[代码](https://github.com/langchain-ai/langchain)]
+
+---
+
+### 约束工程框架分类
+
+**分类依据：**我们根据**约束工程完备性**对智能体系统进行分类——即每个系统实现了六个组件（E, T, C, S, L, V）中的哪些——从而区分全栈约束工程（全部六个组件）与专用框架（部分实现）。
+
+**重要性：**先前的分类法按应用领域（编程、网页导航、具身AI）或模型架构（单智能体、多智能体）对智能体进行分类。这些分类无法解释为何使用相似模型的系统会产生不同的可靠性结果。我们以约束工程框架为中心的分类法揭示，生产级系统收敛于完整的ETCSLV实现，而研究原型往往仅实现2-3个组件。
+
+**核心发现：**没有任何智能体框架能够在不实现**全部六个治理组件**的情况下达到生产可靠性。缺少L组件（生命周期钩子）的系统无法执行安全策略。缺少V组件（评估接口）的系统无法调试故障。缺少S组件（状态持久化）的系统无法从崩溃中恢复。
+
+#### 全栈式约束工程
+
+- <u>PRISM/OpenClaw</u>: **"OpenClaw PRISM: A Zero-Fork, Defense-in-Depth Runtime Security Layer for Tool-Augmented LLM Agents"**. *Li.* arXiv 2026. [[论文](https://arxiv.org/abs/2603.11853)]
+- <u>AIOS</u>: **"AIOS: LLM Agent Operating System"**. *Mei et al.* COLM 2025. [[论文](https://arxiv.org/abs/2403.16971)] [[代码](https://github.com/agiresearch/AIOS)]
+- <u>OpenHands</u>: **"OpenHands: An Open Platform for AI Software Developers as Generalist Agents"**. *Wang et al.* ICLR 2025. [[论文](https://arxiv.org/abs/2407.16741)] [[代码](https://github.com/All-Hands-AI/OpenHands)]
+- <u>SWE-agent</u>: **"SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering"**. *Yang et al.* NeurIPS 2024. [[论文](https://arxiv.org/abs/2405.15793)] [[代码](https://github.com/SWE-agent/SWE-agent)]
+- <u>HAL</u>: **"Holistic Agent Leaderboard: The Missing Infrastructure for AI Agent Evaluation"**. *Kapoor et al.* ICLR 2026. [[论文](https://arxiv.org/abs/2510.11977)]
+
+#### 多智能体约束工程框架
+
+- <u>MetaGPT</u>: **"MetaGPT: Meta Programming for a Multi-Agent Collaborative Framework"**. *Hong et al.* ICLR 2024. [[论文](https://arxiv.org/abs/2308.00352)] [[代码](https://github.com/geekan/MetaGPT)]
+- <u>AutoGen</u>: **"AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation"**. *Wu et al.* arXiv 2023. [[论文](https://arxiv.org/abs/2308.08155)] [[代码](https://github.com/microsoft/autogen)]
+- <u>ChatDev</u>: **"ChatDev: Communicative Agents for Software Development"**. *Qian et al.* ACL 2024. [[论文](https://arxiv.org/abs/2307.07924)] [[代码](https://github.com/OpenBMB/ChatDev)]
+- <u>CAMEL</u>: **"CAMEL: Communicative Agents for 'Mind' Exploration of Large Language Model Society"**. *Li et al.* NeurIPS 2023. [[论文](https://arxiv.org/abs/2303.17760)] [[代码](https://github.com/camel-ai/camel)]
+
+#### 框架与模块
+
+- <u>LangGraph</u>: **"LangGraph: Build Resilient Language Agents as Graphs"**. *LangChain team.* GitHub 2024. [[代码](https://github.com/langchain-ai/langgraph)]
+- <u>MemGPT</u>: **"MemGPT: Towards LLMs as Operating Systems"**. *Packer et al.* NeurIPS 2023. [[论文](https://arxiv.org/abs/2310.08560)] [[代码](https://github.com/cpacker/MemGPT)]
+- <u>Voyager</u>: **"Voyager: An Open-Ended Embodied Agent with Large Language Models"**. *Wang et al.* arXiv 2023. [[论文](https://arxiv.org/abs/2305.16291)] [[代码](https://github.com/MineDojo/Voyager)]
+- <u>Reflexion</u>: **"Reflexion: Language Agents with Verbal Reinforcement Learning"**. *Shinn et al.* NeurIPS 2023. [[论文](https://arxiv.org/abs/2303.11366)] [[代码](https://github.com/noahshinn/reflexion)]
+- <u>Generative Agents</u>: **"Generative Agents: Interactive Simulacra of Human Behavior"**. *Park et al.* UIST 2023. [[论文](https://arxiv.org/abs/2304.03442)] [[代码](https://github.com/joonspk-research/generative_agents)]
+- <u>LangChain</u>: **"LangChain: Building Applications with LLMs through Composability"**. *Chase et al.* GitHub 2022. [[代码](https://github.com/langchain-ai/langchain)]
+- <u>LlamaIndex</u>: **"LlamaIndex: A Data Framework for LLM Applications"**. *Liu et al.* GitHub 2022. [[代码](https://github.com/run-llama/llama_index)]
+- <u>DeerFlow</u>: **"DeerFlow: Distributed Workflow Engine for LLM Agents"**. *GitHub 2024.* [[代码](https://github.com/modelscope/DeerFlow)]
+- <u>DeepAgents</u>: **"DeepAgents: Multi-Agent Framework for Deep Learning"**. *GitHub 2024.* [[代码](https://github.com/deepagents/deepagents)]
+
+#### 评估基础设施
+
+- <u>AgentBench</u>: **"AgentBench: Evaluating LLMs as Agents"**. *Liu et al.* ICLR 2024. [[论文](https://arxiv.org/abs/2308.03688)] [[代码](https://github.com/THUDM/AgentBench)]
+- <u>SWE-bench</u>: **"SWE-bench: Can Language Models Resolve Real-World GitHub Issues?"**. *Jimenez et al.* ICLR 2024. [[论文](https://arxiv.org/abs/2310.06770)] [[代码](https://github.com/swebench/SWE-bench)]
+- <u>OSWorld</u>: **"OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments"**. *Xie et al.* NeurIPS 2024. [[论文](https://arxiv.org/abs/2404.07972)] [[代码](https://github.com/xlang-ai/OSWorld)]
+- <u>WebArena</u>: **"WebArena: A Realistic Web Environment for Building Autonomous Agents"**. *Zhou et al.* ICLR 2024. [[论文](https://arxiv.org/abs/2307.13854)] [[代码](https://github.com/web-arena-x/webarena)]
+- <u>GAIA</u>: **"GAIA: A Benchmark for General AI Assistants"**. *Mialon et al.* ICLR 2024. [[论文](https://arxiv.org/abs/2311.12983)]
+- <u>Mind2Web</u>: **"Mind2Web: Towards a Generalist Agent for the Web"**. *Deng et al.* NeurIPS 2023. [[论文](https://arxiv.org/abs/2306.06070)]
+- <u>AgentBoard</u>: **"AgentBoard: An Analytical Evaluation Board of Multi-Turn LLM Agents"**. *Ma et al.* NeurIPS 2024. [[论文](https://arxiv.org/abs/2401.13178)]
+- <u>InterCode</u>: **"InterCode: Standardizing and Benchmarking Interactive Coding with Execution Feedback"**. *Yang et al.* NeurIPS 2023. [[论文](https://arxiv.org/abs/2306.14898)]
+
+---
+
+### 技术挑战
+
+**范围界定：**我们识别了九项开放研究问题，当前约束工程基础设施针对这些问题仅提供了部分解决方案，而缺乏生产级、可推广的框架。每项挑战都代表了学术原型所展示的成果与已部署系统所需条件之间的差距。
+
+**评估标准：**对于每项挑战，我们评估：(1) **学术覆盖度**——同行评审解决方案（如有）；(2) **生产成熟度**——是否有任何已部署系统在规模上解决了该问题；(3) **阻塞严重性**——未解决的问题是阻止部署还是仅限制性能。
+
+**横切主题：**大多数挑战源于**组合问题**——单个组件（安全沙箱、记忆存储、评估指标）可以独立工作，但将它们组合成连贯的执行器会浮现涌现性故障。形式化验证、可移植性测试和协议互操作性都需要当前研究尚未解决的组合式推理。
+
+#### 安全与沙箱
+
+**挑战：**智能体沙箱必须防止恶意工具使用，同时允许合法的系统访问——这比传统应用沙箱更困难，因为智能体**有意地**与敏感资源（文件系统、网络API、shell命令）交互。
+
+**技术现状：**SandboxEscapeBench 显示前沿模型能够逃逸容器；PRISM 展示了零分叉纵深防御是可能的，但需要操作系统级监测。尚不存在适用于所有智能体用例且无需人工逐任务配置的通用沙箱。
+
+**开放问题：**组合式安全属性（证明智能体+约束工程框架+环境满足安全规范）仍未解决。当前防御是反应式的（检测逃逸）而非预防式的（证明逃逸不可能）。
+
+> **关键数据：**SandboxEscapeBench——前沿LLM达到**15–35%容器逃逸率**；PRISM——10个钩子的零分叉运行时将逃逸降至接近零，开销<5ms。
+
+*（安全与沙箱相关论文列表与英文版相同，此处省略重复）*
+
+#### 评估与基准测试
+
+> **关键数据：**HAL 统一了**20,000+次评估**，将数周的评估压缩至数小时；OSWorld 报告了**28%假阴性率**的自动化评估；METR 发现通过基准测试的PR人工合并率**低24.2个百分点**，差距以9.6pp/年速度扩大。
+
+*（评估与基准测试相关论文列表与英文版相同，此处省略重复）*
+
+#### 协议标准化
+
+> **关键数据：**MCP（工具↔约束工程）：2–15ms延迟；A2A（智能体↔智能体）：50–200ms；ACP（意图级，IBM）——三种协议服务于互补角色。
+
+*（协议标准化相关论文列表与英文版相同，此处省略重复）*
+
+#### 运行时上下文管理
+
+> **关键数据：**SkillsBench——精选技能注入产生**+16.2pp**改进；"迷失在中间"效应已被记录；长上下文模型将问题从*保留*转移至*显著性*。
+
+*（运行时上下文管理相关论文列表与英文版相同，此处省略重复）*
+
+#### 工具使用与注册
+
+> **关键数据：**Vercel 发现移除**80%工具**比任何模型升级都更有帮助；Schema First (Sigdel & Baral, 2026) —— 一项受控的初步研究表明，基于 schema 的工具契约可以减少**接口层面**的误用（如格式校验失败），但无法减少**语义层面**的误用（即格式正确但任务不当的调用），且各实验条件下端到端任务成功率均为零，表明仅靠接口设计不足以保证工具调用的可靠性；CodeAct 在**17/17项Mint基准**上表现优异，**回合数减少20%**。
+
+*（工具使用与注册相关论文列表与英文版相同，此处省略重复）*
+
+#### 记忆架构
+
+> **关键数据：**Mem0 实现了相比完整上下文**90% token减少**；Zep 时序知识：**+18.5% QA准确率**；Agent Workflow Memory：在Mind2Web上**+14.9%**。六种架构模式：平面缓冲区→层次化→情节式→语义式→过程式→图式。
+
+*（记忆架构相关论文列表与英文版相同，此处省略重复）*
+
+#### 规划与推理
+
+> **关键数据：**SWE-agent ACI 研究表明，接口设计超越模型能力成为性能的首要决定因素。LATS 将MCTS与语言模型反馈集成用于状态空间搜索。
+
+*（规划与推理相关论文列表与英文版相同，此处省略重复）*
+
+#### 多智能体协同
+
+> **关键数据：**AgencyBench——智能体在原生SDK约束工程上达到**48.4%成功率**，而在独立约束工程框架上显著更低，展示了约束工程-智能体的紧密耦合。对抗性多智能体设置中的拜占庭容错仍是开放问题。
+
+<p align="center">
+  <img src="assets/multi_agent_topology.png" width="680" alt="多智能体协同拓扑"/>
+</p>
+
+*（多智能体协同相关论文列表与英文版相同，此处省略重复）*
+
+#### 计算经济学
+
+> **关键数据：**OpenRouter 报告**每周13T tokens**（2026年2月），每4周翻倍；AgencyBench 测量平均**1M tokens/任务**；预计2027年智能体计算量增长1000倍；AIOS 通过恰当的智能体调度实现**2.1×吞吐加速**。
+
+*（计算经济学相关论文列表与英文版相同，此处省略重复）*
+
+---
+
+### 新兴主题
+
+*（新兴主题论文列表与英文版相同，此处省略重复）*
+
+#### 相关综述
+
+*（相关综述论文列表与英文版相同，此处省略重复）*
+
+#### 业界实践报告与洞察
+
+> 来自 Stripe、OpenAI、Cursor、METR 及其他前沿实践者的生产部署经验。
+
+*（业界实践报告列表与英文版相同，此处省略重复）*
+
+---
+
+### 未来方向
+
+综述中识别的八个开放研究方向（无精选论文列表——这些是前瞻性方向）：
+
+- **形式化约束工程规范语言** — 用于指定和验证 H=(E,T,C,S,L,V) 组件的领域特定语言（DSL）
+- **跨约束工程框架基准测试套件** — 跨不兼容执行器生态系统的可移植性测试
+- **安全分类与威胁模型** — 将OWASP Top 10扩展至智能体执行器攻击面
+- **协议互操作性（MCP/A2A）** — 桥接工具级和智能体级协议
+- **长时域评估方法论** — 在多会话、多天任务下��会退化的指标
+- **约束工程感知微调** — 训练模型使其感知执行环境
+- **记忆接口标准化** — 跨平面式、情节式和图式存储的可移植记忆API
+- **约束工程透明性规范** — 运行时决策的可审计性和可解释性
+
+---
+
+## 引用
+
+参见本 README 顶部的 BibTeX 引用格式。
+
+---
+
+## 更新日志
+
+| 版本 | 日期 | 变更 |
+|------|------|------|
+|  v1 | 2026-04-03 | 首次预印本 |
+|  v2 | 2026-04-07 | 仓库更新 |
+
+---
+
+<p align="center">
+  <i>† 表示预印本，尚未经同行评审。</i><br>
+  <i>本综述正在积极开发中；完整手稿将于近期发布。</i><br>
+  <i>由 Qianyu Meng 与 Liyi Chen 维护。欢迎提交缺失论文或更新链接的 PR。</i>
+</p>

license

@@ -0,0 +1,395 @@
+Attribution 4.0 International
+
+=======================================================================
+
+Creative Commons Corporation ("Creative Commons") is not a law firm and
+does not provide legal services or legal advice. Distribution of
+Creative Commons public licenses does not create a lawyer-client or
+other relationship. Creative Commons makes its licenses and related
+information available on an "as-is" basis. Creative Commons gives no
+warranties regarding its licenses, any material licensed under their
+terms and conditions, or any related information. Creative Commons
+disclaims all liability for damages resulting from their use to the
+fullest extent possible.
+
+Using Creative Commons Public Licenses
+
+Creative Commons public licenses provide a standard set of terms and
+conditions that creators and other rights holders may use to share
+original works of authorship and other material subject to copyright
+and certain other rights specified in the public license below. The
+following considerations are for informational purposes only, are not
+exhaustive, and do not form part of our licenses.
+
+     Considerations for licensors: Our public licenses are
+     intended for use by those authorized to give the public
+     permission to use material in ways otherwise restricted by
+     copyright and certain other rights. Our licenses are
+     irrevocable. Licensors should read and understand the terms
+     and conditions of the license they choose before applying it.
+     Licensors should also secure all rights necessary before
+     applying our licenses so that the public can reuse the
+     material as expected. Licensors should clearly mark any
+     material not subject to the license. This includes other CC-
+     licensed material, or material used under an exception or
+     limitation to copyright. More considerations for licensors:
+    wiki.creativecommons.org/Considerations_for_licensors
+
+     Considerations for the public: By using one of our public
+     licenses, a licensor grants the public permission to use the
+     licensed material under specified terms and conditions. If
+     the licensor's permission is not necessary for any reason--for
+     example, because of any applicable exception or limitation to
+     copyright--then that use is not regulated by the license. Our
+     licenses grant only permissions under copyright and certain
+     other rights that a licensor has authority to grant. Use of
+     the licensed material may still be restricted for other
+     reasons, including because others have copyright or other
+     rights in the material. A licensor may make special requests,
+     such as asking that all changes be marked or described.
+     Although not required by our licenses, you are encouraged to
+     respect those requests where reasonable. More considerations
+     for the public:
+    wiki.creativecommons.org/Considerations_for_licensees
+
+=======================================================================
+
+Creative Commons Attribution 4.0 International Public License
+
+By exercising the Licensed Rights (defined below), You accept and agree
+to be bound by the terms and conditions of this Creative Commons
+Attribution 4.0 International Public License ("Public License"). To the
+extent this Public License may be interpreted as a contract, You are
+granted the Licensed Rights in consideration of Your acceptance of
+these terms and conditions, and the Licensor grants You such rights in
+consideration of benefits the Licensor receives from making the
+Licensed Material available under these terms and conditions.
+
+
+Section 1 -- Definitions.
+
+  a. Adapted Material means material subject to Copyright and Similar
+     Rights that is derived from or based upon the Licensed Material
+     and in which the Licensed Material is translated, altered,
+     arranged, transformed, or otherwise modified in a manner requiring
+     permission under the Copyright and Similar Rights held by the
+     Licensor. For purposes of this Public License, where the Licensed
+     Material is a musical work, performance, or sound recording,
+     Adapted Material is always produced where the Licensed Material is
+     synched in timed relation with a moving image.
+
+  b. Adapter's License means the license You apply to Your Copyright
+     and Similar Rights in Your contributions to Adapted Material in
+     accordance with the terms and conditions of this Public License.
+
+  c. Copyright and Similar Rights means copyright and/or similar rights
+     closely related to copyright including, without limitation,
+     performance, broadcast, sound recording, and Sui Generis Database
+     Rights, without regard to how the rights are labeled or
+     categorized. For purposes of this Public License, the rights
+     specified in Section 2(b)(1)-(2) are not Copyright and Similar
+     Rights.
+
+  d. Effective Technological Measures means those measures that, in the
+     absence of proper authority, may not be circumvented under laws
+     fulfilling obligations under Article 11 of the WIPO Copyright
+     Treaty adopted on December 20, 1996, and/or similar international
+     agreements.
+
+  e. Exceptions and Limitations means fair use, fair dealing, and/or
+     any other exception or limitation to Copyright and Similar Rights
+     that applies to Your use of the Licensed Material.
+
+  f. Licensed Material means the artistic or literary work, database,
+     or other material to which the Licensor applied this Public
+     License.
+
+  g. Licensed Rights means the rights granted to You subject to the
+     terms and conditions of this Public License, which are limited to
+     all Copyright and Similar Rights that apply to Your use of the
+     Licensed Material and that the Licensor has authority to license.
+
+  h. Licensor means the individual(s) or entity(ies) granting rights
+     under this Public License.
+
+  i. Share means to provide material to the public by any means or
+     process that requires permission under the Licensed Rights, such
+     as reproduction, public display, public performance, distribution,
+     dissemination, communication, or importation, and to make material
+     available to the public including in ways that members of the
+     public may access the material from a place and at a time
+     individually chosen by them.
+
+  j. Sui Generis Database Rights means rights other than copyright
+     resulting from Directive 96/9/EC of the European Parliament and of
+     the Council of 11 March 1996 on the legal protection of databases,
+     as amended and/or succeeded, as well as other essentially
+     equivalent rights anywhere in the world.
+
+  k. You means the individual or entity exercising the Licensed Rights
+     under this Public License. Your has a corresponding meaning.
+
+
+Section 2 -- Scope.
+
+  a. License grant.
+
+       1. Subject to the terms and conditions of this Public License,
+          the Licensor hereby grants You a worldwide, royalty-free,
+          non-sublicensable, non-exclusive, irrevocable license to
+          exercise the Licensed Rights in the Licensed Material to:
+
+            a. reproduce and Share the Licensed Material, in whole or
+               in part; and
+
+            b. produce, reproduce, and Share Adapted Material.
+
+       2. Exceptions and Limitations. For the avoidance of doubt, where
+          Exceptions and Limitations apply to Your use, this Public
+          License does not apply, and You do not need to comply with
+          its terms and conditions.
+
+       3. Term. The term of this Public License is specified in Section
+          6(a).
+
+       4. Media and formats; technical modifications allowed. The
+          Licensor authorizes You to exercise the Licensed Rights in
+          all media and formats whether now known or hereafter created,
+          and to make technical modifications necessary to do so. The
+          Licensor waives and/or agrees not to assert any right or
+          authority to forbid You from making technical modifications
+          necessary to exercise the Licensed Rights, including
+          technical modifications necessary to circumvent Effective
+          Technological Measures. For purposes of this Public License,
+          simply making modifications authorized by this Section 2(a)
+          (4) never produces Adapted Material.
+
+       5. Downstream recipients.
+
+            a. Offer from the Licensor -- Licensed Material. Every
+               recipient of the Licensed Material automatically
+               receives an offer from the Licensor to exercise the
+               Licensed Rights under the terms and conditions of this
+               Public License.
+
+            b. No downstream restrictions. You may not offer or impose
+               any additional or different terms or conditions on, or
+               apply any Effective Technological Measures to, the
+               Licensed Material if doing so restricts exercise of the
+               Licensed Rights by any recipient of the Licensed
+               Material.
+
+       6. No endorsement. Nothing in this Public License constitutes or
+          may be construed as permission to assert or imply that You
+          are, or that Your use of the Licensed Material is, connected
+          with, or sponsored, endorsed, or granted official status by,
+          the Licensor or others designated to receive attribution as
+          provided in Section 3(a)(1)(A)(i).
+
+  b. Other rights.
+
+       1. Moral rights, such as the right of integrity, are not
+          licensed under this Public License, nor are publicity,
+          privacy, and/or other similar personality rights; however, to
+          the extent possible, the Licensor waives and/or agrees not to
+          assert any such rights held by the Licensor to the limited
+          extent necessary to allow You to exercise the Licensed
+          Rights, but not otherwise.
+
+       2. Patent and trademark rights are not licensed under this
+          Public License.
+
+       3. To the extent possible, the Licensor waives any right to
+          collect royalties from You for the exercise of the Licensed
+          Rights, whether directly or through a collecting society
+          under any voluntary or waivable statutory or compulsory
+          licensing scheme. In all other cases the Licensor expressly
+          reserves any right to collect such royalties.
+
+
+Section 3 -- License Conditions.
+
+Your exercise of the Licensed Rights is expressly made subject to the
+following conditions.
+
+  a. Attribution.
+
+       1. If You Share the Licensed Material (including in modified
+          form), You must:
+
+            a. retain the following if it is supplied by the Licensor
+               with the Licensed Material:
+
+                 i. identification of the creator(s) of the Licensed
+                    Material and any others designated to receive
+                    attribution, in any reasonable manner requested by
+                    the Licensor (including by pseudonym if
+                    designated);
+
+                ii. a copyright notice;
+
+               iii. a notice that refers to this Public License;
+
+                iv. a notice that refers to the disclaimer of
+                    warranties;
+
+                 v. a URI or hyperlink to the Licensed Material to the
+                    extent reasonably practicable;
+
+            b. indicate if You modified the Licensed Material and
+               retain an indication of any previous modifications; and
+
+            c. indicate the Licensed Material is licensed under this
+               Public License, and include the text of, or the URI or
+               hyperlink to, this Public License.
+
+       2. You may satisfy the conditions in Section 3(a)(1) in any
+          reasonable manner based on the medium, means, and context in
+          which You Share the Licensed Material. For example, it may be
+          reasonable to satisfy the conditions by providing a URI or
+          hyperlink to a resource that includes the required
+          information.
+
+       3. If requested by the Licensor, You must remove any of the
+          information required by Section 3(a)(1)(A) to the extent
+          reasonably practicable.
+
+       4. If You Share Adapted Material You produce, the Adapter's
+          License You apply must not prevent recipients of the Adapted
+          Material from complying with this Public License.
+
+
+Section 4 -- Sui Generis Database Rights.
+
+Where the Licensed Rights include Sui Generis Database Rights that
+apply to Your use of the Licensed Material:
+
+  a. for the avoidance of doubt, Section 2(a)(1) grants You the right
+     to extract, reuse, reproduce, and Share all or a substantial
+     portion of the contents of the database;
+
+  b. if You include all or a substantial portion of the database
+     contents in a database in which You have Sui Generis Database
+     Rights, then the database in which You have Sui Generis Database
+     Rights (but not its individual contents) is Adapted Material; and
+
+  c. You must comply with the conditions in Section 3(a) if You Share
+     all or a substantial portion of the contents of the database.
+
+For the avoidance of doubt, this Section 4 supplements and does not
+replace Your obligations under this Public License where the Licensed
+Rights include other Copyright and Similar Rights.
+
+
+Section 5 -- Disclaimer of Warranties and Limitation of Liability.
+
+  a. UNLESS OTHERWISE SEPARATELY UNDERTAKEN BY THE LICENSOR, TO THE
+     EXTENT POSSIBLE, THE LICENSOR OFFERS THE LICENSED MATERIAL AS-IS
+     AND AS-AVAILABLE, AND MAKES NO REPRESENTATIONS OR WARRANTIES OF
+     ANY KIND CONCERNING THE LICENSED MATERIAL, WHETHER EXPRESS,
+     IMPLIED, STATUTORY, OR OTHER. THIS INCLUDES, WITHOUT LIMITATION,
+     WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR
+     PURPOSE, NON-INFRINGEMENT, ABSENCE OF LATENT OR OTHER DEFECTS,
+     ACCURACY, OR THE PRESENCE OR ABSENCE OF ERRORS, WHETHER OR NOT
+     KNOWN OR DISCOVERABLE. WHERE DISCLAIMERS OF WARRANTIES ARE NOT
+     ALLOWED IN FULL OR IN PART, THIS DISCLAIMER MAY NOT APPLY TO YOU.
+
+  b. TO THE EXTENT POSSIBLE, IN NO EVENT WILL THE LICENSOR BE LIABLE
+     TO YOU ON ANY LEGAL THEORY (INCLUDING, WITHOUT LIMITATION,
+     NEGLIGENCE) OR OTHERWISE FOR ANY DIRECT, SPECIAL, INDIRECT,
+     INCIDENTAL, CONSEQUENTIAL, PUNITIVE, EXEMPLARY, OR OTHER LOSSES,
+     COSTS, EXPENSES, OR DAMAGES ARISING OUT OF THIS PUBLIC LICENSE OR
+     USE OF THE LICENSED MATERIAL, EVEN IF THE LICENSOR HAS BEEN
+     ADVISED OF THE POSSIBILITY OF SUCH LOSSES, COSTS, EXPENSES, OR
+     DAMAGES. WHERE A LIMITATION OF LIABILITY IS NOT ALLOWED IN FULL OR
+     IN PART, THIS LIMITATION MAY NOT APPLY TO YOU.
+
+  c. The disclaimer of warranties and limitation of liability provided
+     above shall be interpreted in a manner that, to the extent
+     possible, most closely approximates an absolute disclaimer and
+     waiver of all liability.
+
+
+Section 6 -- Term and Termination.
+
+  a. This Public License applies for the term of the Copyright and
+     Similar Rights licensed here. However, if You fail to comply with
+     this Public License, then Your rights under this Public License
+     terminate automatically.
+
+  b. Where Your right to use the Licensed Material has terminated under
+     Section 6(a), it reinstates:
+
+       1. automatically as of the date the violation is cured, provided
+          it is cured within 30 days of Your discovery of the
+          violation; or
+
+       2. upon express reinstatement by the Licensor.
+
+     For the avoidance of doubt, this Section 6(b) does not affect any
+     right the Licensor may have to seek remedies for Your violations
+     of this Public License.
+
+  c. For the avoidance of doubt, the Licensor may also offer the
+     Licensed Material under separate terms or conditions or stop
+     distributing the Licensed Material at any time; however, doing so
+     will not terminate this Public License.
+
+  d. Sections 1, 5, 6, 7, and 8 survive termination of this Public
+     License.
+
+
+Section 7 -- Other Terms and Conditions.
+
+  a. The Licensor shall not be bound by any additional or different
+     terms or conditions communicated by You unless expressly agreed.
+
+  b. Any arrangements, understandings, or agreements regarding the
+     Licensed Material not stated herein are separate from and
+     independent of the terms and conditions of this Public License.
+
+
+Section 8 -- Interpretation.
+
+  a. For the avoidance of doubt, this Public License does not, and
+     shall not be interpreted to, reduce, limit, restrict, or impose
+     conditions on any use of the Licensed Material that could lawfully
+     be made without permission under this Public License.
+
+  b. To the extent possible, if any provision of this Public License is
+     deemed unenforceable, it shall be automatically reformed to the
+     minimum extent necessary to make it enforceable. If the provision
+     cannot be reformed, it shall be severed from this Public License
+     without affecting the enforceability of the remaining terms and
+     conditions.
+
+  c. No term or condition of this Public License will be waived and no
+     failure to comply consented to unless expressly agreed to by the
+     Licensor.
+
+  d. Nothing in this Public License constitutes or may be interpreted
+     as a limitation upon, or waiver of, any privileges and immunities
+     that apply to the Licensor or You, including from the legal
+     processes of any jurisdiction or authority.
+
+
+=======================================================================
+
+Creative Commons is not a party to its public
+licenses. Notwithstanding, Creative Commons may elect to apply one of
+its public licenses to material it publishes and in those instances
+will be considered the “Licensor.” The text of the Creative Commons
+public licenses is dedicated to the public domain under the CC0 Public
+Domain Dedication. Except for the limited purpose of indicating that
+material is shared under a Creative Commons public license or as
+otherwise permitted by the Creative Commons policies published at
+creativecommons.org/policies, Creative Commons does not authorize the
+use of the trademark "Creative Commons" or any other trademark or logo
+of Creative Commons without its prior written consent including,
+without limitation, in connection with any unauthorized modifications
+to any of its public licenses or any other arrangements,
+understandings, or agreements concerning use of licensed material. For
+the avoidance of doubt, this paragraph does not form part of the
+public licenses.
+
+Creative Commons may be contacted at creativecommons.org.