{"title":"LLM-Agent-UMF: LLM-based Agent Unified Modeling Framework for Seamless Integration of Multi Active/Passive Core-Agents","authors":"Amine B. Hassouna, Hana Chaari, Ines Belhaj","doi":"arxiv-2409.11393","DOIUrl":null,"url":null,"abstract":"The integration of tools in LLM-based agents overcame the difficulties of\nstandalone LLMs and traditional agents' limited capabilities. However, the\nconjunction of these technologies and the proposed enhancements in several\nstate-of-the-art works followed a non-unified software architecture resulting\nin a lack of modularity. Indeed, they focused mainly on functionalities and\noverlooked the definition of the component's boundaries within the agent. This\ncaused terminological and architectural ambiguities between researchers which\nwe addressed in this paper by proposing a unified framework that establishes a\nclear foundation for LLM-based agents' development from both functional and\nsoftware architectural perspectives. Our framework, LLM-Agent-UMF (LLM-based Agent Unified Modeling Framework),\nclearly distinguishes between the different components of an agent, setting\nLLMs, and tools apart from a newly introduced element: the core-agent, playing\nthe role of the central coordinator of the agent which comprises five modules:\nplanning, memory, profile, action, and security, the latter often neglected in\nprevious works. Differences in the internal structure of core-agents led us to\nclassify them into a taxonomy of passive and active types. Based on this, we\nproposed different multi-core agent architectures combining unique\ncharacteristics of various individual agents. For evaluation purposes, we applied this framework to a selection of\nstate-of-the-art agents, thereby demonstrating its alignment with their\nfunctionalities and clarifying the overlooked architectural aspects. Moreover,\nwe thoroughly assessed four of our proposed architectures by integrating\ndistinctive agents into hybrid active/passive core-agents' systems. This\nanalysis provided clear insights into potential improvements and highlighted\nthe challenges involved in the combination of specific agents.","PeriodicalId":501278,"journal":{"name":"arXiv - CS - Software Engineering","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Software Engineering","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.11393","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
The integration of tools in LLM-based agents overcame the difficulties of
standalone LLMs and traditional agents' limited capabilities. However, the
conjunction of these technologies and the proposed enhancements in several
state-of-the-art works followed a non-unified software architecture resulting
in a lack of modularity. Indeed, they focused mainly on functionalities and
overlooked the definition of the component's boundaries within the agent. This
caused terminological and architectural ambiguities between researchers which
we addressed in this paper by proposing a unified framework that establishes a
clear foundation for LLM-based agents' development from both functional and
software architectural perspectives. Our framework, LLM-Agent-UMF (LLM-based Agent Unified Modeling Framework),
clearly distinguishes between the different components of an agent, setting
LLMs, and tools apart from a newly introduced element: the core-agent, playing
the role of the central coordinator of the agent which comprises five modules:
planning, memory, profile, action, and security, the latter often neglected in
previous works. Differences in the internal structure of core-agents led us to
classify them into a taxonomy of passive and active types. Based on this, we
proposed different multi-core agent architectures combining unique
characteristics of various individual agents. For evaluation purposes, we applied this framework to a selection of
state-of-the-art agents, thereby demonstrating its alignment with their
functionalities and clarifying the overlooked architectural aspects. Moreover,
we thoroughly assessed four of our proposed architectures by integrating
distinctive agents into hybrid active/passive core-agents' systems. This
analysis provided clear insights into potential improvements and highlighted
the challenges involved in the combination of specific agents.