This is a short, hands-on course on building with coding agents: how to structure agent loops, when autonomy helps vs hurts, how tools and memory are actually used, and where things tend to break in real projects.
The course focuses on:
* Practical agent loop patterns (plan → act → reflect and variants)
* Tool use, context management, and guardrails
* Common failure modes (hallucinations, drift, runaway autonomy)
* Tradeoffs between control, reliability, and speed
It’s aimed at engineers who want to go beyond toy examples and actually ship things with coding agents. Not a framework, not a product — just structured lessons based on real-world patterns.
Feedback and critiques welcome, especially from people already building agents in production.