Context Engineering 2.0: The Context of Context Engineering
Qishuo Hua, Lyumanshan Ye, Dayuan Fu, Yang Xiao, Xiaojie Cai, Yunze Wu, Jifan Lin, Junfei Wang, Pengfei Liu
TL;DR
This paper reframes context engineering as a long-running discipline that bridges human intent and machine understanding, arguing that context should be treated as a holistic construct shaped by entropy reduction rather than a narrow set of prompts or dialogue history. It advances a formal framework for context definition and a four-stage evolutionary model (primitive computation, agent-centric intelligence, human-level intelligence, and speculative superhuman intelligence) to organize past work and future directions. The authors trace a historical arc from 1.0 ubiquitous computing foundations to 2.0 agent-centric approaches, then articulate design considerations across context collection, management, and usage, including textual and multimodal processing, layered memory, subagents, and cross-system sharing. They also outline practical implications through applications in CLI, deep research, and BCIs, and discuss challenges such as lifelong context preservation, scalability, evaluation, and semantic operating-system-like architectures. Overall, the work provides a conceptual foundation and roadmap for systematic, scalable context engineering in AI systems with potential broad impact on long-horizon reasoning and human–machine collaboration.
Abstract
Karl Marx once wrote that ``the human essence is the ensemble of social relations'', suggesting that individuals are not isolated entities but are fundamentally shaped by their interactions with other entities, within which contexts play a constitutive and essential role. With the advent of computers and artificial intelligence, these contexts are no longer limited to purely human--human interactions: human--machine interactions are included as well. Then a central question emerges: How can machines better understand our situations and purposes? To address this challenge, researchers have recently introduced the concept of context engineering. Although it is often regarded as a recent innovation of the agent era, we argue that related practices can be traced back more than twenty years. Since the early 1990s, the field has evolved through distinct historical phases, each shaped by the intelligence level of machines: from early human--computer interaction frameworks built around primitive computers, to today's human--agent interaction paradigms driven by intelligent agents, and potentially to human--level or superhuman intelligence in the future. In this paper, we situate context engineering, provide a systematic definition, outline its historical and conceptual landscape, and examine key design considerations for practice. By addressing these questions, we aim to offer a conceptual foundation for context engineering and sketch its promising future. This paper is a stepping stone for a broader community effort toward systematic context engineering in AI systems.