Implementation:Langchain ai Langgraph Langgraph Dev Command

Property	Value
API	`langgraph dev` CLI command
Type	External Tool Doc
Source	`libs/cli/langgraph_cli/cli.py`
Library	langgraph-cli
Related Workflow	CLI_Deployment

Overview

The `langgraph dev` command runs a local LangGraph API server in development mode with hot reloading and debugging support. It executes entirely in-process (no Docker required), loading graph definitions from the project configuration file and serving them via the LangGraph API. The command requires the `langgraph-cli[inmem]` extra to be installed, which provides the `langgraph_api` package containing the server runtime.

Description

Command Registration

The `dev` command is registered as a Click subcommand of the `cli` group at `libs/cli/langgraph_cli/cli.py:L680-768`. It is decorated with `@log_command` for analytics tracking.

Execution Flow

Import check: Attempts to import `run_server` from `langgraph_api.cli`. If unavailable, raises a `UsageError` with installation instructions.
Configuration loading: Calls `validate_config_file` on the specified configuration path to load and validate the configuration.
Node.js check: Rejects configurations with `node_version` set (JS in-memory server is not supported in the Python CLI).
Path setup: Adds the current working directory and all local dependency directories to `sys.path` for import resolution.
Server startup: Calls `run_server` from `langgraph_api.cli` with the extracted configuration parameters.

Command Options

Option	Type	Default	Description
`--host`	`str`	`127.0.0.1`	Network interface to bind to
`--port`	`int`	`2024`	Port number to bind to
`--no-reload`	flag	`False`	Disable automatic code reloading
`--config`	`Path`	`langgraph.json`	Path to configuration file
`--n-jobs-per-worker`	`int`	`None` (10)	Maximum concurrent jobs per worker
`--no-browser`	flag	`False`	Skip opening browser on startup
`--debug-port`	`int`	`None`	Port for remote debugging via `debugpy`
`--wait-for-client`	flag	`False`	Wait for debugger client before starting
`--studio-url`	`str`	`None`	Custom LangGraph Studio URL
`--allow-blocking`	flag	`False`	Suppress errors for synchronous I/O in async code
`--tunnel`	flag	`False`	Expose local server via Cloudflare tunnel
`--server-log-level`	`str`	`WARNING`	Log level for the API server

Parameters Passed to run_server

The `dev` command extracts the following from the validated configuration and passes them to `run_server`:

`graphs`: The graph registry dictionary.
`env`: Environment variables (dict or file path).
`store`: Store configuration for long-term memory.
`auth`: Authentication configuration.
`http`: HTTP server configuration.
`ui`: UI component definitions.
`ui_config`: UI configuration.
`webhooks`: Webhook configuration.

Code Reference

Source Location

Item	File	Line
`dev` function	`libs/cli/langgraph_cli/cli.py`	L685-768
Click options	`libs/cli/langgraph_cli/cli.py`	L609-683

Signature

@cli.command("dev", help="Run LangGraph API server in development mode with hot reloading and debugging support")
@log_command
def dev(
    host: str,
    port: int,
    no_reload: bool,
    config: str,
    n_jobs_per_worker: int | None,
    no_browser: bool,
    debug_port: int | None,
    wait_for_client: bool,
    studio_url: str | None,
    allow_blocking: bool,
    tunnel: bool,
    server_log_level: str,
):
    """CLI entrypoint for running the LangGraph API server."""
    ...

Import

This is a CLI command, not a Python import. It is invoked from the command line:

langgraph dev [OPTIONS]

The required package is installed with:

pip install -U "langgraph-cli[inmem]"

I/O Contract

Direction	Name	Type	Description
Input	CLI options	various	Command-line flags and arguments
Input	config file	JSON file	`langgraph.json` (or specified via `--config`)
Output	HTTP server	network	API server listening on `host:port`
Output	Browser	side effect	Opens LangGraph Studio in default browser (unless `--no-browser`)
Raises	`click.UsageError`		If `langgraph-api` is not installed, configuration is invalid, or Node.js graphs are specified

Usage Examples

Basic Development Startup

# Start development server with defaults (localhost:2024)
langgraph dev

Custom Port and Config

# Serve on port 8000 with a custom config
langgraph dev --port 8000 --config ./my_project/langgraph.json

Remote Debugging with VS Code

# Start with debug port and wait for VS Code to attach
langgraph dev --debug-port 5678 --wait-for-client

Then in VS Code `launch.json`:

{
    "name": "Attach to LangGraph",
    "type": "debugpy",
    "request": "attach",
    "connect": {"host": "localhost", "port": 5678}
}

Tunnel for Remote Studio Access

# Expose local server via Cloudflare tunnel
langgraph dev --tunnel --no-browser

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment