framework-selection

SKILL.md

Framework selection guide for LangChain, LangGraph, and Deep Agents layered architecture.

Layered frameworks where LangChain provides foundation primitives, LangGraph adds orchestration and control flow, and Deep Agents adds planning, memory, file management, and skill delegation

Decision table guides framework choice based on task complexity: LangChain for single-purpose agents, LangGraph for custom control flow and loops, Deep Agents for multi-step planning and persistent sessions

Deep Agents includes six built-in middleware modules (todo lists, filesystem, subagents, skills, memory, human-in-the-loop) that can be configured or extended

Frameworks can be mixed in the same project; LangGraph graphs can be registered as subagents within Deep Agents, and LangChain primitives work across all layers

┌─────────────────────────────────────────┐
│              Deep Agents                │  ← highest level: batteries included
│   (planning, memory, skills, files)     │
├─────────────────────────────────────────┤
│               LangGraph                 │  ← orchestration: graphs, loops, state
│    (nodes, edges, state, persistence)   │
├─────────────────────────────────────────┤
│               LangChain                 │  ← foundation: models, tools, chains
│      (models, tools, prompts, RAG)      │
└─────────────────────────────────────────┘

Picking a higher layer does not cut you off from lower layers — you can use LangGraph graphs inside Deep Agents, and LangChain primitives inside both.