Experience Compression Spectrum: Unifying Memory, Skills, and Rules in LLM Agents

2026-04-17

AI Summary

This work addresses the challenges of context inflation, retrieval latency, and computational overhead in long-term, multi-session deployments of large language model agents caused by accumulating experience. It proposes the “Experience Compression Spectrum,” a theoretical framework that unifies memory, skills, and rules as distinct levels along a knowledge compression axis. Through bibliometric analysis, system mapping, and theoretical abstraction, the study identifies a pervasive “missing diagonal” deficiency—namely, the lack of cross-level adaptive compression capabilities—and highlights the neglect of knowledge lifecycle management in existing systems. Quantitative results demonstrate substantial compression ratios across levels (5–20× for memory, 50–500× for skills, and >1000× for rules), revealing a trade-off wherein higher compression enhances transferability at the expense of specificity. The paper concludes with design principles for full-spectrum scalable agents, bridging the divide between memory-centric and skill-discovery research communities.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Multiagent Systems: Modeling other Agents
• Data Mining & Knowledge Management: Data Compression

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

As LLM agents scale to long-horizon, multi-session deployments, efficiently managing accumulated experience becomes a critical bottleneck. Agent memory systems and agent skill discovery both address this challenge -- extracting reusable knowledge from interaction traces -- yet a citation analysis of 1,136 references across 22 primary papers reveals a cross-community citation rate below 1%. We propose the \emph{Experience Compression Spectrum}, a unifying framework that positions memory, skills, and rules as points along a single axis of increasing compression (5--20$\times$ for episodic memory, 50--500$\times$ for procedural skills, 1,000$\times$+ for declarative rules), directly reducing context consumption, retrieval latency, and compute overhead. Mapping 20+ systems onto this spectrum reveals that every system operates at a fixed, predetermined compression level -- none supports adaptive cross-level compression, a gap we term the \emph{missing diagonal}. We further show that specialization alone is insufficient -- both communities independently solve shared sub-problems without exchanging solutions -- that evaluation methods are tightly coupled to compression levels, that transferability increases with compression at the cost of specificity, and that knowledge lifecycle management remains largely neglected. We articulate open problems and design principles for scalable, full-spectrum agent learning systems.

Problem

• experience compression
• LLM agents
• memory systems
• skill discovery
• knowledge lifecycle

Innovation

• Experience Compression Spectrum
• LLM agents
• adaptive compression
• knowledge lifecycle
• unified framework