LangChain CLI Cheatsheet

Q: How do I get started with LangChain CLI?

Install LangChain CLI with `pip install langchain-cli`. Create a new app with `langchain app new my-app`. Add templates with `langchain app add rag-chroma`. Start the server with `langchain serve` and visit `http://localhost:8000/docs` for the API docs.

Q: What are LangChain templates?

Templates are pre-built LangServe applications for common patterns (RAG with Chroma, conversational retrieval, SQL agent, etc.). List available templates with `langchain template list`. Add them to your app with `langchain app add `.

Q: How is LangGraph different from LangChain?

LangChain is for building chains and RAG pipelines. LangGraph is for building agent-based, cyclic workflows with state management. Use LangGraph when you need multi-step agents with tool calling, reflection loops, or human-in-the-loop. Install with `pip install langgraph`.

Q: How do I deploy a LangGraph agent?

Run `langgraph dev` for local development. For production, use `langgraph build -t my-agent:latest` to build a Docker image, then `langgraph up` to start with Docker Compose. The agent serves an OpenAI-compatible API at `/chat`.

Q: How do I enable LangSmith tracing?

Set `export LANGCHAIN_TRACING_V2=true` and `export LANGCHAIN_API_KEY=ls-xxx`. Get your API key from smith.langchain.com. All LangChain/LangGraph calls will be logged to the LangSmith dashboard for debugging and evaluation.

The LangChain CLI ecosystem provides two tools: `langchain` for application scaffolding and LangServe deployment, and `langgraph` for agent workflow development. Both support template-based quick-start, REST API deployment, and Docker packaging for production.

Updated: 2026-07-20·24 commands

Quick Start

``bash # Install CLI pip install langchain-cli

# Create a new app langchain app new my-app

# Add a RAG template langchain app add rag-chroma

# Create environment file with API keys echo 'OPENAI_API_KEY=sk-xxx' > .env

# Start LangServe langchain serve --port 8000

# Visit the docs open http://localhost:8000/docs`

`LangGraph`

For agent workflows, install and use LangGraph separately:

`bash pip install langgraph

# Start dev server langgraph dev

# Build for production langgraph build -t my-agent:v1

# Deploy with Docker Compose langgraph up`

`Templates`

Templates provide pre-built architectures: RAG, conversational retrieval, SQL agents, document analysis, and more. Browse and add them via the CLI.

`LangServe`

LangServe wraps LangChain chains as REST APIs automatically. Each template has its own endpoint at /chain-name, plus a Playground UI at /chain-name/playground` for interactive testing.

LangSmith

Observability and debugging platform. Enable tracing via environment variables to log all LLM calls, chains, and agent steps for inspection and evaluation.

Startup & Modes(16)

Command	Level	Example
`langchain --help` Show LangChain CLI help	Basic	`langchain --help`
`langchain app new` Create a new LangChain application	Basic	`langchain app new my-app`
`langchain app new --package` Create app from a template package	Intermediate	`langchain app new my-app --package rag-chroma`
`langchain app add` Add a package (template) to existing app	Intermediate	`langchain app add rag-chroma`
`pip install langchain-cli` Install LangChain CLI	Basic	`pip install langchain-cli`
`pip install langchain` Install LangChain core library	Basic	`pip install langchain`
`pip install langchain-community` Install LangChain community integrations	Basic	`pip install langchain-community`
`pip install langgraph` Install LangGraph framework	Intermediate	`pip install langgraph`
`langgraph build` Build LangGraph app for deployment	Intermediate	`langgraph build -t my-agent:latest`
`langgraph build -t` Specify Docker image tag	Intermediate	`langgraph build -t my-agent:v1`
`langgraph test` Run LangGraph tests	Intermediate	`langgraph test`
`langchain template list` List all available template packages	Basic	`langchain template list`
`pip install langchain-postgres` Install LangChain PostgreSQL integration	Intermediate	`pip install langchain-postgres`
`pip install langchain-chroma` Install LangChain Chroma integration	Intermediate	`pip install langchain-chroma`
`export LANGCHAIN_API_KEY` Set LangSmith API key	Basic	`export LANGCHAIN_API_KEY=ls-xxx`
`export LANGCHAIN_TRACING_V2=true` Enable LangSmith tracing	Intermediate	`export LANGCHAIN_TRACING_V2=true`

ai-server(8)

Command	Level	Example
`langchain serve` Start LangServe REST API server	Intermediate	`langchain serve`
`langchain serve --port` Specify server port	Basic	`langchain serve --port 8080`
`langchain serve --host` Specify bind host address	Basic	`langchain serve --host 0.0.0.0`
`langgraph dev` Start LangGraph development server	Intermediate	`langgraph dev`
`langgraph up` Start LangGraph app with Docker Compose	Intermediate	`langgraph up`
`npx @langchain/langgraph-cli dev` Start LangGraph dev server via npx	Intermediate	`npx @langchain/langgraph-cli dev`
`curl http://localhost:8000/docs` Access LangServe API documentation	Basic	`curl -s http://localhost:8000/docs`
`curl http://localhost:8000/rag/playground` Access LangServe Playground UI	Basic	`curl -s http://localhost:8000/rag/playground`

FAQ

How do I get started with LangChain CLI?

What are LangChain templates?

How is LangGraph different from LangChain?

How do I deploy a LangGraph agent?

How do I enable LangSmith tracing?

This cheatsheet is compiled from official tool documentation. Last updated: 2026-07-20.