这是为OpenClaw AI代理生态系统精心策划的学术论文、行业报告、数据集和工具合集。

我们调查的配套仓库：OpenClaw 生态系统调查——从平台可扩展性到约束设计。

📄 阅读调查

OpenClaw 生态系统综述——从平台可扩展性到约束设计 可在本仓库获取 PDF 版本：OpenClaw_Survey.pdf。

OpenClaw —— 由 Peter Steinberger（Clawdbot → Moltbot → OpenClaw 创建的开源、自托管的 AI 代理平台，于 2026 年 1 月 29 日发布——在不到四个月的时间内，已生成了 **74 篇学术论文、23 篇基准测试 以及 18+ 份主要行业报告。本仓库采用了我们调查中引入的 PSEA（平台–安全–学会–部署） 分类法，组织了研究领域。

调查论点。 OpenClaw 最好被理解为对开放个人代理生态系统的压力测试。其开放技能、持久内存和始终在线的 Heartbeat 使能力易于扩展，但同样的开放性也带来了治理、安全、社交和部署等问题。

Awesome OpenClaw Research

A curated collection of academic papers, industry reports, datasets, and tools for the OpenClaw AI agent ecosystem.

Companion repository for our survey: A Survey of the OpenClaw Ecosystem — From Platform Extensibility to Constraint Design.

📄 Read the Survey

A Survey of the OpenClaw Ecosystem — From Platform Extensibility to Constraint Design is available as a PDF in this repository: OpenClaw_Survey.pdf.

OpenClaw — the open-source, self-hosted AI agent platform created by Peter Steinberger (Clawdbot → Moltbot → OpenClaw, January 29, 2026) — has generated 74 academic papers, 23 benchmarks, and 18+ major industry reports in under four months. This repository organizes the research landscape using the PSEA (Platform–Security–Societies–Deployment) taxonomy introduced in our survey.

Thesis of the survey. OpenClaw is best understood as a stress test for open personal-agent ecosystems. Its open Skills, persistent Memory, and always-on Heartbeat make capability easy to extend, but the same openness creates governance, security, social, and deployment problems. The literature converges on one recurring tradeoff: extensibility accelerates capability growth, but trustworthy use requires explicit constraints on Skills, Memory, autonomy, domain actions, and evaluation. The repository is organized to make this tradeoff visible at every level.

---

Statistics at a Glance