One repo, five agent frameworks, six weeks of homework

What it does

This is the companion repository for a course on agentic AI engineering. It collects Jupyter notebooks, setup scripts, and troubleshooting guides that walk students through building autonomous agents across five different frameworks. Think of it as a curated workbook rather than a standalone library.

The interesting bit

Instead of committing to one ecosystem, the course treats agent frameworks like a buffet—sample OpenAI’s SDK, CrewAI, LangGraph, AutoGen, and MCP in a single progression. The README is refreshingly honest about the rough edges, warning Windows users about Build Tools gotchas and CrewAI unicode bugs before they hit them.

Key highlights

• Covers five frameworks in one curriculum: OpenAI Agents SDK, CrewAI, LangGraph, AutoGen, and MCP.
• Includes platform-specific setup guides and a troubleshooting notebook for common environment pitfalls.
• Provides Ollama and DeepSeek alternatives for students trying to avoid API costs.
• Structured as a six-week progression with video resources and community support channels.

Caveats

• This is course material, not a reusable package; expect glue code and homework rather than importable tools.
• Several framework-specific workarounds are already documented (CrewAI version pinning, Windows unicode fixes), suggesting the ecosystem is still settling.

Verdict

Worth bookmarking if you want a structured, comparative introduction to agent frameworks and don’t mind following along with a course. Skip it if you’re looking for a single, production-ready agent library to drop into your stack.