Agentic Architecture

Use when deciding how an LLM-backed system should be structured — whether it needs an agent at all or a deterministic workflow, which orchestration pattern fits (chain, router, parallel fan-out, orchestrator-worker, evaluator-optimizer, supervisor, hierarchical), how many agents, how state and durability work, where tools and human approval gates belong, how loops terminate within a budget, which framework class to adopt and when, and how to evaluate a trajectory rather than a single answer.

joogy06 f489ecb 9.7 KB Updated

File contents

joogy06/agent-foundry/tree/main/skills/agentic-architecture commit f489ecb087

Frequently asked questions

npx skillmds@latest add joogy06/agentic-architecture