framework-selection

Choosing between Deep Agents, LangGraph, and LangChain depends on task complexity: use LangChain for simple single-purpose tool usage, LangGraph for complex control flow, and Deep Agents for open-ended multi-step needs. This skill provides structured decision guidance to map requirements to the correct framework layer.

Use LangGraph instead of LangChain when your AI agent workflow requires complex control flow. LangChain is better suited for simple single-purpose tool usage, while LangGraph handles state management and intricate routing logic for multi-step agent tasks.

Yes, Deep Agents are designed for open-ended multi-step needs. If your AI project involves unpredictable task execution paths requiring high autonomy, this framework layer provides the necessary architecture compared to simpler LangChain or LangGraph setups.

The best way to bootstrap a LangChain or LangGraph project is to load this framework-selection skill first at project startup. It aligns tooling with architecture goals and provides practical workflows to identify the correct framework layer upfront before development.

Yes, you can mix framework layers in the same project. This skill provides clear framework profiles and recommended next skills to invoke, enabling smooth mixing of Deep Agents, LangGraph, and LangChain layers to align tooling with your architecture goals.