Description
AI Agents From Scratch: Tools, Memory, MCP and Multi-Agent is a course on building AI agents published by Udemy Online Academy. Rather than relying solely on high-level frameworks, this course focuses on the core components that allow agents to reason, maintain context, use external tools, communicate with services, and collaborate with other agents. Learners will learn about agent architecture, tool invocation, memory systems, Model Context Protocol (MCP), task scheduling, agent loops, context management, and multi-agent coordination, while developing practical agent workflows. Rather than starting with a pre-built framework and treating the agent as a black box, this course breaks everything down to the basics so you know exactly what’s going on under the hood. You build the core tool-calling loop yourself — designing schemas, validating arguments, and handling errors and exceptions that inevitably occur when an agent actually runs on real tools and data, not just a clean demo.
Building on these foundations, you examine task scheduling, agent loops, and context management as interconnected topics, not separate topics, and see how they interact when an agent must perform multiple tasks without losing sight of its purpose. The course concludes by extending single-agent skills to multi-agent coordination, showing how individual agents can be designed to communicate with each other and collaborate on common tasks. Throughout the course, the focus remains entirely practical: every concept is tied to code you write and run yourself, so by the end, you’ll have a real, ground-up understanding of how to build real AI agents, not just how to configure someone else’s framework.
What you will learn in AI Agents From Scratch: Tools, Memory, MCP and Multi-Agent:
- Build a production AI agent from scratch: the tool-calling loop, schema design, argument validation, and error handling that survives a real run
- Give agents real memory with file-based persistence, RAG and chunking, plus retrieval that stays fresh instead of quietly going stale across sessions
- Write your own MCP server from scratch: expose tools and schemas, handle auth, connect it to Claude Code, and debug connection and schema failures
- and …
Course specifications
Publisher: Udemy
Instructors: Ferbin Richard
Language: English
Level: Introductory to Advanced
Number of Lessons: 68
Duration: 5 hours and 50 minutes
Course topics
AI Agents From Scratch: Tools, Memory, MCP and Multi-Agent Prerequisites
Comfortable writing basic Python. If you can write a function and run a script from the terminal, you have everything you need to follow along.
No API key required. The course builds the agent loop against a mock model, so every demo runs on your own machine for free.
No prior experience with AI agents, MCP or LLM tooling. The course starts at what separates an agent from a chatbot and builds up from there.
Pictures
AI Agents From Scratch: Tools, Memory, MCP and Multi-Agent introduction video
Installation guide
After Extract, watch with your favorite Player.
Subtitle: None
Quality: 1080p
Downloadly link
Rapidgator link
File password (s): www.downloadly.ir
Size
2.2 GB