LLM-Agent+: A Modular Framework for Intelligent Agents with Reasoning Trace Compression and Tool-Augmented Memory

We present LLM-Agent+, a modular and extensible framework for building intelligent agents poweredby Large Language Models (LLMs). Designed for both research and real-world deployment, LLM-Agent+integrates natural language understanding (NLU), a dual-layer memory system, a chain-of-thought (CoT)reasoning engine, and a standardized tool interface—enabling flexible and scalable agent development.A key innovation is the Reasoning Trace Compression (RTC) mechanism, which dynamically condensesintermediate reasoning steps to improve memory efficiency, reduce prompt overhead, and enhanceinterpretability in long-context tasks. Unlike existing frameworks, RTC adapts to context windowconstraints, making LLM-Agent+ particularly effective in multi-step reasoning scenarios. The frameworksupports both command-line and web-based interfaces, emphasizing high modularity to facilitaterapid prototyping of alternative memory and reasoning strategies. We evaluated LLM-Agent+ across arange of tasks, including software debugging, multi-step planning, and research synthesis. The resultsdemonstrate competitive performance with a significantly reduced memory footprint, improved tasksuccess rates through enhanced reasoning transparency, and up to 40% reduction in token usagecompared to baseline methods. The dual-layer memory system further contributes to effective longcontextmanagement. To promote reproducibility and community collaboration, the full source codehas been released under a permissive open-source license. LLM-Agent+ bridges modular reasoningwith efficient context management, offering a unified platform for developing scalable and transparentLLM-based agents. Its balance of performance, flexibility, and interpretability positions it as a strongfoundation for future research in intelligent agent systems.Keywords: Intelligent Agents, Reasoning Trace Compression (RTC).