TopoClaw: A Human-Centric and Topology-Aware Agent Operating System

2026-05-15

AI Summary

This work proposes a human-centric agent operating system to address the limitations of existing agent-centric designs, which struggle to support distributed, highly collaborative, and permission-sensitive human-agent teamwork. The system models the user ecosystem as a coupled structure of device topology and social relationship graphs, treating agents as socially situated digital twins. It enables accountable proactive autonomy through three core mechanisms: cross-device action scheduling, cross-user identity provenance, and cross-context permission governance. Integrating device topology modeling, social graph analysis, distributed policy enforcement, and role-aware access control, the architecture provides an open-source, engineering-oriented reference implementation encompassing runtime management, cross-device execution, collaboration protocols, and a robust security model, thereby effectively supporting complex human-agent collaborative deployments.

Abstract

Large language models (LLMs) have evolved AI assistants into autonomous reasoning engines that maintain context, invoke tools, and pursue long-horizon tasks. This has spurred Agent Operating Systems (Agent OS) as kernel-like layers for lifecycle management, memory, scheduling, and access control. Yet most designs remain agent-centric, treating the OS as a single-host runtime for internal reasoning and tool use, leaving open how autonomous actions integrate with distributed, collaborative, permission-sensitive workflows. TopoClaw is an open-source, human-centric, topology-aware Agent OS modeling the user's ecosystem as two coupled structures: a physical device topology of heterogeneous surfaces and a social relationship topology of shared spaces, teams, and delegated roles. It unifies device operation, messaging, and skills around accountable cross-boundary execution, with three core contributions: (1) cross-device action placement, decoupling intent from actuation and routing distributed actions across the device cluster based on hardware affordances and user context; (2) cross-user identity attribution, treating agents as socially situated"Digital Twins"that coordinate in multi-user spaces while preserving provenance, role-aware permissions, and human accountability; (3) cross-context authority governance, pairing broad capability with distributed, context-aware policy enforcement across physical and social trust boundaries to bound proactive autonomy at the OS layer. This report presents TopoClaw as an engineering-oriented reference architecture, covering its design principles, runtime, cross-device execution, collaboration mechanisms, security model, and deployment outlook.

Problem

• Agent Operating System
• human-centric computing
• topology-aware
• cross-boundary execution
• autonomous agents

Innovation

• Agent Operating System
• Topology-Aware
• Human-Centric AI
• Cross-Device Execution
• Digital Twin