liveBook · Manning

Agent orchestration framework

As explained in Context Engineering

An agent orchestration framework is a software framework for controlling multi-step behavior, state, tool use, and collaboration among agents.

An agent orchestration framework is a developer library for coordinating agent behavior across multiple steps. It manages how state, tool observations, and memories evolve as an agent plans, acts, revises results, requests human approval, or collaborates with other agents. section 10.1 section 10.4

The framework governs what each step sees, how state changes, how tool outputs enter subsequent model calls, how agents exchange information, and when execution stops. Some frameworks provide machinery for teams to define the agent loop and its boundaries; others provide a pre-assembled harness with established defaults. section 10.4 section 10.7

Long-running coding agent

A long-running coding agent needs explicit state, approvals, and execution control across steps. By contrast, a single retrieval-augmented generation endpoint does not necessarily need multi-agent orchestration; a simpler pipeline may be more reliable. section 10.2

Choose and operate an orchestration framework around the system’s requirements:

  • Define context sources, retrieval needs, tool boundaries, memory, state, evaluation, and operational constraints before selecting a tool. section 10.2
  • Inspect the context sent to the model, tool outputs reintroduced into it, and state retained across steps. Watch for context drift across agents and tool outputs. section 10.2
  • Plan for traces, evaluations, logs, cost monitoring, policy checks, rollback, and safe secret handling. Keep prompts, retrieval configuration, tool definitions, and evaluation data outside framework-specific formats where possible. section 10.2
Table 10.1 A taxonomy of AI frameworks and platforms for context engineering
Category Primary role Examples

AI application frameworks

Developer libraries for building LLM applications with different context sources and model integrations.

LangChain, LlamaIndex, Haystack, Spring AI, LangChain4j, Pydantic AI, DSPy, AI SDK

Agent orchestration frameworks

Build agents and coordinate multi-step execution, planning, tool use, agent state, human-in-the-loop control, and multi-agent workflows. Some supply the machinery for building the agent loop, and others deliver it already assembled.

LangGraph, CrewAI, Agent Development Kit, Microsoft Agent Framework, Agno, Embabel, Parlant, DeepAgents, Claude Agent SDK

AI application platforms

Integrated environments for visual design, deployment, monitoring, team workflows, and operational control.

Dify, LangFlow, Flowise, Zapier, n8n, OpenClaw, Temporal, Amazon Bedrock AgentCore, Gemini Enterprise Agent Platform

Compare orchestration frameworks with application frameworks and integrated platforms by role and representative tools. section 10.1

How Context Engineering covers it

How deep it goes
Contrasts developer-defined loops with pre-assembled harnesses and gives selection and operational criteria without a code-level implementation.
Examples
Contrasts a long-running coding agent with a single RAG endpoint and names LangGraph, CrewAI, DeepAgents, and Claude Agent SDK.
What you can do after
Evaluate orchestration frameworks against architecture, context visibility, operational controls, and portability requirements.
How it connects
Connects orchestration to agent loops, state management, tool execution, human-in-the-loop control, and multi-agent workflows.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Context Engineering. The section links point to the parts of the book it draws on.