#patterns

Six-hour agent runs are now real. The harness — checkpoints, durable state, recovery — matters more than the model. A field guide to the long-running pattern.

When one agent isn't enough, hierarchy is usually the first answer. Supervisor agents, worker pools, and the patterns that keep multi-agent systems coherent at scale.

The three-role pattern that powered Manus, then LangChain Deep Agents, then half the production agents shipping in early 2026. Why it works, when it doesn't, and how to actually build one.

Past ~30 tools, agent reliability falls off a cliff. Past ~100, it's chaos. Here's the actual engineering — RAG-over-tools, semantic routing, dynamic loading, and namespacing — that production teams ship to stay sane.

The orchestrator-worker pattern that took over agent design in late 2025 is the same pattern that took over backend design in 2014. The wins are real. So are the failure modes.

The three control-flow patterns that turn a toy LangGraph agent into a production system — bounded cycles, parallel fan-out/merge branches, and interrupt-resume human approval gates.