An Empirical Study of Agent Skills for Healthcare: Practice, Gaps, and Governance

2026-05-04

AI Summary

This study addresses the challenges of skill reuse and misaligned risk assessment in deploying medical AI agents across institutions. It introduces “agent skills” as a programmable intermediate layer for medical AI adaptation, leveraging the ClawHub platform to curate 557 healthcare-related skills. Through expert annotation across ten dimensions and empirical analysis, the work systematically characterizes these skills in terms of functionality, deployment contexts, autonomy, and safety. Findings reveal a predominant focus on patient-facing workflow automation, with insufficient coverage of core clinical tasks and uneven representation across the healthcare lifecycle. Moreover, conventional technical risk metrics fail to capture real-world clinical risks. These insights expose critical blind spots in current benchmarking and governance frameworks, offering a novel paradigm for evaluating the transferability and safety of medical AI systems.

Abstract

Healthcare automation is shaped by local procedures and organizational constraints, so agent capabilities rarely transfer unchanged across settings. Agent skills, self-contained directories that package reusable procedures for AI agents, are emerging as a procedural layer for adapting healthcare agents across diverse healthcare settings. We present the first empirical analysis of healthcare agent skills, drawing on 557 healthcare-related skills filtered from 58,159 public skills on ClawHub and annotated along ten dimensions covering function, deployment context, autonomy, and safety. We find that public healthcare skills emphasize patient-facing workflow automation and monitoring rather than the diagnostic and treatment-oriented tasks foregrounded in healthcare-agent research; coverage of the healthcare lifecycle and specialized clinical inputs remains uneven; and general technical risk does not reliably capture clinical risk. These findings position healthcare skills as a procedural layer not yet addressed by current benchmarks and risk frameworks.

Problem

• healthcare automation
• agent skills
• clinical risk
• workflow automation
• procedural layer

Innovation

• agent skills
• healthcare automation
• procedural layer
• clinical risk
• empirical analysis