Skip to main content

06 · Agents & Orchestration

Multi-step autonomous systems. Concepts first, frameworks second — understand the patterns before picking the tools.

StepTopicOne-linerStatus
1AI agent fundamentalsWhat makes something an agent: perception, reasoning, action, memory🔴
2Tool / function callingHow LLMs invoke external functions and interpret their results🔴
3ReAct patternReason → Act → Observe loop — the foundational agent architecture🔴
4Chain of ThoughtPrompting the model to reason step-by-step before answering🔴
5Plan & executeSeparating planning from execution for complex multi-step tasks🔴
6Human-in-the-loopPatterns for routing agent decisions to humans for approval🔴
7MCP — Model Context ProtocolStandardized protocol for agents to discover and call tools🔴
8Agent SDKSDK-level primitives for building agents (Anthropic, OpenAI, Google)🔴
9LangGraphGraph-based agent orchestration with explicit state and control flow🔴
10LangChain & LlamaIndexThe two dominant agent/RAG frameworks and when to use each🔴
11Multi-agent systemsNetworks of specialized agents collaborating on shared tasks🔴
12Handoff — agent to agentPatterns for transferring control between agents cleanly🔴
13Tool registryCentralised catalog of available tools with schemas and permissions🔴
14Idempotent tool callsMaking tool calls safe to retry without side effects🔴
15A2A protocolGoogle's agent-to-agent communication protocol (2025)🔴

← Previous section: 05 · Retrieval & Memory | Next section → 07 · Safety, Alignment & Governance