When the model is allowed to do things, not only to answer.
Python
Claude
OpenAI
FastAPIAn agent is a model inside a harness: a loop, tools, memory and limits. This track is about the harness — how an agent loop runs step by step, designing tools a model can use correctly, managing a context that keeps growing, multi-agent systems, and the part most often skipped: guardrails. An agent allowed to call an API is a system that can be wrong with real consequences.
Read a free lesson: What Makes Something an AgentEvery module and lesson in this track. Lessons marked free are open to everyone; the rest open once you own the track.
Every lesson in this track, with summariesWhat an agent is, what a harness is, and when a problem actually needs an agent — and when it does not.
Context, permissions, sandboxing, errors and tracing — everything that makes an agent trustworthy outside a demo.
Connecting agents to APIs, databases and files — with MCP as the standard interface.
Orchestrators and workers, subagents and context isolation — when splitting the work helps, and when it hurts.
Task-level evals, trajectory analysis, and reliability and cost metrics.
Human-in-the-loop, security, and a capstone: an agent that completes a real task safely.
The real one, from a lesson — it runs in your browser. No account needed.
This track is part of a larger path. Buy the whole path as a bundle and every track in it costs 40% less.
