AI Agent Series

🇬🇧 English

🇫🇷 Français

A from-scratch walkthrough of building an AI agent, one stage at a time. Each post pairs with a folder under agents/ in the repo source, so you can read the explanation and the code side by side.

Posts

• Intro: Building an AI Agent From Scratch

• Agent-001: The Echo Loop

• Agent-002: The LLM Call

• Agent-003: Conversation History

• Agent-004: The System Prompt

• Agent-005: The First Tool

• Agent-006: Closing the Loop

• Agent-007: A Toolbox, Not a Tool

• Agent-008: Sturdiness Before Substance

• Agent-009: Giving the Agent Knowledge

• Agent-010: An Agent that manages the Context

• Agent-011: Subagent - An Agent as a Tool

• Agent-012: Evals and Tracing - Trust, but Verify

• Agent-013: The Capstone - Everything at Once, Nothing New