AgentHub: A Research Agenda for Agent Sharing Infrastructure

TL;DR

The paper addresses the fragmentation of agent-sharing infrastructure by proposing AgentHub, a registry-centric research agenda to enable reliable, scalable, and auditable sharing of LLM-based agents. It defines key agent-specific requirements—capability clarity, lifecycle transparency, interoperability, openness and governance, trust and security, and discovery/workflow integration—and outlines canonical manifests, evidence pipelines, SBOM-style dependencies, and governance considerations. Through lessons from traditional software/model registries and analyses of emerging agent ecosystems, it outlines concrete design directions, evaluation metrics, and a roadmap for building a unified ecosystem that supports safe composition and reuse of agent components. The work aims to catalyze practical infrastructure that mirrors成熟 software registries while addressing the unique autonomy and evolution of agents, ultimately enabling trusted, seamless agent sharing across environments.

Abstract

LLM-based agents are rapidly proliferating, yet the infrastructure for discovering, evaluating, and governing them remains fragmented compared to mature ecosystems like software package registries (e.g., npm) and model hubs (e.g., Hugging Face). Recent research and engineering works have begun to consider the requisite infrastructure, but so far they focus narrowly -- on distribution, naming, or protocol negotiation. However, considering broader software engineering requirements would improve open-source distribution and ease reuse. We therefore propose AgentHub, a research agenda for agent sharing. By framing the key challenges of capability clarity, lifecycle transparency, interoperability, governance, security, and workflow integration, AgentHub charts a community-wide agenda for building reliable and scalable agent ecosystems. Our vision is a future where agents can be shared, trusted, and composed as seamlessly as today's software libraries.

Figures (2)

• Figure 1: Conceptual view of AgentHub, illustrating how publishers, identities, and agent protocols might interact.
• Figure 2: Research agenda for AgentHub, showing how six requirements (§3.2) encounter specific challenges, motivating research directions toward objectives of reproducibility, portability, resilience, and fair discovery.