chapter ten

10 AI frameworks for context engineering

 

This chapter covers

  • A taxonomy organizing the ecosystem into application frameworks, orchestration frameworks, and platforms
  • Architectural criteria, trade-offs, and decision paths for selecting a framework
  • Practical context abstractions and patterns across leading open-source and enterprise frameworks
  • Design surfaces, durable workflow engines, and cloud-managed agent runtimes

The previous chapters examined the building blocks of context engineering, including instructions, external knowledge, tools, memory, prompts, context management, orchestration, evaluation, observability, governance, and operations. Few teams implement all of them from scratch. Most rely on frameworks and platforms that package those capabilities as reusable abstractions, and that choice decides how visible and how controllable the context lifecycle becomes. This chapter organizes the ecosystem into three groups, sets out the criteria for choosing among them, and then reviews representative projects in each group from a context engineering perspective.

10.1 Understanding the framework landscape

10.2 Choosing a framework for your AI application

10.3 AI application frameworks

10.3.1 LangChain

10.3.2 LlamaIndex

10.3.3 Haystack

10.3.4 Spring AI

10.3.5 LangChain4j

10.3.6 Pydantic AI

10.3.7 DSPy

10.3.8 AI SDK

10.4 Agent orchestration frameworks

10.4.1 LangGraph

10.4.2 CrewAI

10.4.3 Agent Development Kit

10.4.4 Microsoft Agent Framework

10.4.5 Agno

10.4.6 Embabel

10.4.7 Parlant

10.4.8 DeepAgents

10.4.9 Claude Agent SDK

10.5 AI application platforms

10.5.1 Dify

10.5.2 LangFlow

10.5.3 Flowise

10.5.4 Zapier