AgenticRAGModels

A Survey of Context Engineering for Large Language Models

by Mei et al. (arXiv)

AdvancedPaperFree~long-form survey

1,400+ papers distilled into a taxonomy for engineering everything that fills an LLM's context.

Start LearningReviewed July 8, 2026

Overview

A Survey of Context Engineering for Large Language Models (arXiv:2507.13334, submitted July 17, 2025) by Lingrui Mei and colleagues analyzes 1,411 research papers to build a unified taxonomy of context engineering. It decomposes the discipline into foundational components — context retrieval and generation, context processing, and context management — and shows how they compose into system implementations including retrieval-augmented generation (RAG), memory systems, tool-integrated reasoning, and multi-agent architectures. A central finding is a 'fundamental asymmetry': modern models are strong at understanding complex contexts but comparatively weak at generating equally sophisticated long-form outputs, an insight the authors argue should guide real-world system design. The paper serves as a foundational reference and reading map for practitioners moving beyond ad-hoc prompting toward disciplined context design.

At a Glance

Topic
Agentic
Level
Advanced
Format
Paper
Cost
Free
Duration
~long-form survey
Provider
Mei et al. (arXiv)
Hands-on
No
Certificate
None

What You’ll Learn

  • A precise taxonomy of context engineering: retrieval/generation, processing, and management
  • How RAG, memory, tool-integrated reasoning, and multi-agent systems fit a shared framework
  • Design trade-offs for assembling and managing what fills an LLM's context window
  • The 'understanding vs. generation' asymmetry and what it implies for system design
  • A curated map into 1,400+ underlying papers for deeper study

Highlights

  • Synthesizes over 1,411 papers into one coherent taxonomy
  • Covers the full pipeline from retrieval through multi-agent orchestration
  • A go-to reference framing the shift from prompt engineering to context engineering

Who It’s For

Best For

  • Engineers designing RAG, memory, or multi-agent LLM systems
  • Researchers surveying the context-engineering landscape
  • Tech leads defining a principled approach to context design

Prerequisites

  • Solid understanding of LLMs and transformers
  • Familiarity with RAG and agent architectures

FAQ

What is A Survey of Context Engineering for Large Language Models?

A comprehensive 2025 survey that formalizes 'context engineering' — the systematic design of what goes into an LLM's context window — synthesizing over 1,400 papers. It is for engineers and researchers building RAG, memory, tool-use, and multi-agent systems who want a structured map of the field.

Is A Survey of Context Engineering for Large Language Models free?

A Survey of Context Engineering for Large Language Models is free to access.

What level is A Survey of Context Engineering for Large Language Models for?

A Survey of Context Engineering for Large Language Models is aimed at a advanced audience. Recommended background: Solid understanding of LLMs and transformers, Familiarity with RAG and agent architectures.

How long does A Survey of Context Engineering for Large Language Models take?

Expect roughly ~long-form survey. Most learners work through it at their own pace.

What will I learn from A Survey of Context Engineering for Large Language Models?

You'll learn: A precise taxonomy of context engineering: retrieval/generation, processing, and management; How RAG, memory, tool-integrated reasoning, and multi-agent systems fit a shared framework; Design trade-offs for assembling and managing what fills an LLM's context window; The 'understanding vs. generation' asymmetry and what it implies for system design; A curated map into 1,400+ underlying papers for deeper study.

Topics

context-engineeringragllm-agentssurveymemory-systems