Home > Wiki > Toolchain > Langgraph > Concepts > Application Structure

Application Structure

Status: ACTIVE (pulled from docs.langchain.com) Source: https://docs.langchain.com/oss/python/langgraph/application-structure Timestamp: 2026-05-11

A LangGraph app consists of: one or more graphs, a langgraph.json config file, a dependencies file, and an optional .env.

File Structure

my-app/
├── my_agent/
│   ├── utils/
│   │   ├── tools.py      # Tool definitions
│   │   ├── nodes.py      # Node functions
│   │   └── state.py      # State definition
│   ├── __init__.py
│   └── agent.py          # Graph construction
├── .env                   # Environment variables
├── requirements.txt       # Package dependencies
└── langgraph.json         # Configuration file

langgraph.json

{
  "dependencies": ["langchain_openai", "./your_package"],
  "graphs": {
    "my_agent": "./your_package/your_file.py:agent"
  },
  "env": "./.env"
}

Key	Purpose
`dependencies`	Python packages + local packages (`.` or `./path`)
`graphs`	Named graphs with `path/to/file.py:variable` format
`env`	Path to environment file
`dockerfile_lines`	Additional system binaries or libraries

Dependencies

A dependency file (requirements.txt, pyproject.toml, or package.json)
A dependencies key in langgraph.json
Any system-level needs via dockerfile_lines

Graphs

Each graph is identified by a unique name pointing to either: - A compiled graph variable: "./agent.py:agent" - A function that returns a graph: "./agent.py:make_graph"

Environment Variables

For local dev, use the env key in langgraph.json. For production, configure in the deployment environment.