I Built an AI Coding Agent in 200 Lines of Code.
Автор: Jo Van Eyck
Загружено: 2025-08-31
Просмотров: 2881
Ever wondered how AI coding agents like Claude Code, Github Copilot and Codex work? So did I!
So I decided to build an AI coding agent from scratch with a local LLM running on a spare laptop and just 200 lines of code. No black magic, no hidden frameworks. In this video, I’ll show you how agents really work: wiring up a CLI wrapper, tracking conversations, handling tool calls, and running real code through an LLM.
What you’ll learn:
What "agentic AI" really means
How tool calls and recursive inference work
Step-by-step: build your own agent in ~200 LOC
Why frontier models matter for real-world production-grade coding agents
🎵 CTRL ALT DEL by Streambeats/Harris Heller
👩💻 https://github.com/jovaneyck/ai-codin...
📑 https://ampcode.com/how-to-build-an-a...
📑 https://ghuntley.com/agent/
Timestamps
00:00 Introduction
01:55 Agentic AI 101
04:00 Hello world
06:25 Conversations
09:30 Tool calling
15:22 Shell access
17:30 Demo time
19:05 Conclusion
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
