Agent in Depth
This course is for developers, researchers, and product managers interested in large language models (LLMs) and their intelligent agents. The course covers the basic principles of LLMs, the composition of Agents, common protocols and tools, and end-to-end Agent project construction methods. Through a combination of theoretical explanations and practical exercises, the course helps students master the core skills of building intelligent Agent systems from scratch.
Day 1: Agent Theory and System Design
Systematic Understanding from Concept to Architecture
Study Content +
1.1 Agent Concepts and Core Composition
Agent = LLM + Memory + Tools + Knowledge
Explain the composition and operation mechanism of Agent based on LangChain / LangGraph
1.2 Analysis of Mainstream Agent Models
ReAct model: a cycle mechanism of reasoning and action
Function Call: structured tool interaction
MCP (Modular Command Protocol): modular command protocol
A2A (Agent-to-Agent): inter-agent communication protocol
Analysis of applicable scenarios, advantages and limitations of each model
1.3 Agent System Design Methodology
Deconstruct the Agent workflow and draw a flowchart
Methodology for designing an Agentic system from 0:
Demonstrate process modeling based on LangGraph
Learning Objectives +
- โ In-depth understanding of the core composition and working principle of Agent
- โ Master the design ideas and implementation methods of mainstream Agent models
- โ Possess the architectural design capabilities of Agent systems
- โ Proficient in using LangGraph for process modeling
Day 2: Basic Practice and LangGraph Application
Skills Transformation from Theory to Practice
Study Content +
2.1 LangGraph Core Principles
Basic concepts and core modules of LangGraph:
Model a typical Agent flowchart
2.2 Practice: Basic Agent Implementation
ReAct Agent construction
Agent + Tools (Function Call) integration
Agent + MCP:
Learning Objectives +
- โ Proficient in the core concepts and usage of LangGraph
- โ Ability to independently build a basic Agent system
- โ Master the integration and development technology of Agent and tools
- โ Understand and implement Agent integration of MCP protocol
Day 3: Advanced Practice and Multi-Agent Systems
Complex System Design and Collaboration Mechanisms
Study Content +
3.1 Advanced Capabilities Expansion of Agent
Memory & Knowledge Integration:
Human-in-the-Loop Design:
3.2 Multi-Agent System Design and Collaboration Mechanism
Typical architectural patterns of multi-Agent collaboration
Detailed explanation of A2A (Agent-to-Agent) protocol:
3.3 Comprehensive Project Practice
Design a multi-Agent collaboration system from requirements to deployment
Use LangGraph for modeling, debugging and demonstration
3.4 Course Summary and Extension
Course review and frequently asked questions
Suggestions for subsequent learning paths and reference materials
Learning Objectives +
- โ Master the design and implementation of advanced Agent capabilities
- โ Possess the architectural design capabilities of multi-agent systems
- โ Proficient in using A2A protocol for Agent collaboration
- โ Complete an end-to-end multi-Agent system project
- โ Possess the ability to independently design and deploy Agentic applications