Agent Designer

If you're building anything beyond a single AI agent, this gives you the architectural patterns and evaluation frameworks to do it properly. It covers the standard orchestration approaches (supervisor, swarm, pipeline, hierarchical) with actual tradeoffs, plus practical guidance on tool schemas, communication patterns, and guardrails. The evaluation section is especially useful since most people wing agent metrics until things break. Think of it as the missing architecture guide for multi-agent systems. You'll want this when you're past the prototype stage and need to design something that won't collapse under real-world complexity.