This is a production-focused reference for building LLM agents that won't embarrass you in production. It covers the full lifecycle: architecture selection (workflow vs planner vs multi-agent), MCP tool integration, RAG pipelines, guardrails, observability with OpenTelemetry, and deployment patterns. The framework comparison table alone is worth it, comparing LangGraph, Claude Agent SDK, Pydantic AI, and ten others across maturity and use case. What I like is the honesty: it includes ROI calculation templates, build-vs-not decision trees, and explicit kill triggers for doomed projects. No theory, just checklists, templates, and decision trees. Use it when you need to spec an agent system or explain to stakeholders why their autonomous coding agent idea needs guardrails.
npx skills add https://github.com/vasilyu1983/ai-agents-public --skill ai-agents