Title: Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward

Original: arXiv:2602.12430v4 Announce Type: replace-cross Abstract: The transition from monolithic language models to modular, skill-equipped agents marks a defining shift in how large language models (LLMs) are deployed in practice. Rather than encoding all procedural knowledge within model weights, agent skills -- composable packages of instructions, code, and resources that agents load on demand -- enable dynamic capability extension without retraining. It is formalized in a paradigm of progressive disclosure, portable skill definitions, and integration with the Model Context Protocol (MCP). This survey provides a comprehensive treatment of the agent skills landscape, as it has rapidly evolved during the last few months. We organize the field along four axes: (i) architectural foundations, examining the SKILL$.$md specification, progressive context loading, and the complementary roles of skills and MCP; (ii) skill acquisition, covering reinforcement learning with skill libraries, autonomous skill discovery (SEAgent), and compositional skill synthesis; (iii) deployment at scale, including the computer-use agent (CUA) stack, GUI grounding advances, and benchmark progress on OSWorld and SWE-bench; and (iv) security, where recent empirical analyses reveal that 26.1% of community-contributed skills contain vulnerabilities, motivating our proposed Skill Trust and Lifecycle Governance Framework -- a four-tier, gate-based permission model that maps skill provenance to graduated deployment capabilities. We identify seven open challenges -- from cross-platform skill portability to capability-based permission models -- and propose a research agenda for realizing trustworthy, self-improving skill ecosystems. Unlike prior surveys that broadly cover LLM agents or tool use, this work focuses specifically on the emerging skill abstraction layer and its implications for the next generation of agentic systems. Project repo: https://github.com/scienceaix/agentskills

Rewrite: Title: Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward

Original: arXiv:2602.12430v4 Announce Type: replace-cross Abstract: The shift from monolithic language models to modular, skill-equipped agents represents a pivotal change in the practical deployment of large language models (LLMs). Instead of embedding all procedural knowledge directly into model weights, agent skills—modular bundles of instructions, code, and resources loaded on demand—allow for dynamic capability expansion without the need for retraining. This approach is structured around progressive disclosure, portable skill definitions, and integration with the Model Context Protocol (MCP). This survey offers a thorough examination of the rapidly evolving agent skills landscape. We structure the discussion across four key dimensions: (i) architectural foundations, focusing on the SKILL$.$md specification, progressive context loading, and the synergistic relationship between skills and MCP; (ii) skill acquisition, encompassing reinforcement learning with skill libraries, autonomous skill discovery via SEAgent, and compositional skill synthesis; (iii) large-scale deployment, highlighting the computer-use agent (CUA) stack, advancements in GUI grounding, and performance on benchmarks like OSWorld and SWE-bench; and (iv) security, where empirical studies indicate that 26.1% of community-contributed skills harbor vulnerabilities. This finding underscores the need for our proposed Skill Trust and Lifecycle Governance Framework—a four-tier, gate-based permission system that aligns skill provenance with graduated deployment permissions. We outline seven open challenges, ranging from cross-platform skill portability to capability-based permission models, and present a research agenda aimed at fostering trustworthy, self-improving skill ecosystems. Distinct from earlier surveys that broadly address LLM agents or tool use, this study zeroes in on the emerging skill abstraction layer and its impact on future agentic systems. Project repo: https://github.com/scienceaix/agentskills

Source: arXiv Generated at: 2026-06-03 00:00:00 UTC