lfx-nightly 0.3.0.dev36

Langflow Executor - A lightweight CLI tool for executing and serving Langflow AI flows

lfx - Langflow Executor

lfx is a command-line tool for running Langflow workflows. It provides two main commands: serve and run.

Installation

From PyPI (recommended)

# Install globally
uv pip install lfx

# Or run without installing using uvx
uvx lfx serve my_flow.json
uvx lfx run my_flow.json "input"

From source (development)

# Clone and run in workspace
git clone https://github.com/langflow-ai/langflow
cd langflow/src/lfx
uv run lfx serve my_flow.json

Key Features

Pluggable Services

lfx supports a pluggable service architecture that allows you to customize and extend its behavior. You can replace built-in services (storage, telemetry, tracing, etc.) with your own implementations or use Langflow's full-featured services.

📖 See PLUGGABLE_SERVICES.md for details including:

• Quick start guides for CLI users, library developers, and plugin authors
• Service registration via config files, decorators, and entry points
• Creating custom service implementations with dependency injection
• Using full-featured Langflow services in lfx
• Troubleshooting and migration guides

Flattened Component Access

lfx now supports simplified component imports for better developer experience:

Before (old import style):

from lfx.components.agents.agent import AgentComponent
from lfx.components.data.url import URLComponent
from lfx.components.input_output import ChatInput, ChatOutput

Now (new flattened style):

from lfx import components as cp

# Direct access to all components
chat_input = cp.ChatInput()
agent = cp.AgentComponent()
url_component = cp.URLComponent()
chat_output = cp.ChatOutput()

Benefits:

• Simpler imports: One import line instead of multiple deep imports
• Better discovery: All components accessible via cp.ComponentName
• Helpful error messages: Clear guidance when dependencies are missing
• Backward compatible: Traditional imports still work

Commands

lfx serve - Run flows as an API

Serve a Langflow workflow as a REST API.

Important: You must set the LANGFLOW_API_KEY environment variable before running the serve command.

export LANGFLOW_API_KEY=your-secret-key
uv run lfx serve my_flow.json --port 8000

This creates a FastAPI server with your flow available at /flows/{flow_id}/run. The actual flow ID will be displayed when the server starts.

Options:

• --host, -h: Host to bind server (default: 127.0.0.1)
• --port, -p: Port to bind server (default: 8000)
• --verbose, -v: Show diagnostic output
• --env-file: Path to .env file
• --log-level: Set logging level (debug, info, warning, error, critical)
• --check-variables/--no-check-variables: Check global variables for environment compatibility (default: check)

Example:

# Set API key (required)
export LANGFLOW_API_KEY=your-secret-key

# Start server
uv run lfx serve simple_chat.json --host 0.0.0.0 --port 8000

# The server will display the flow ID, e.g.:
# Flow ID: af9edd65-6393-58e2-9ae5-d5f012e714f4

# Call API using the displayed flow ID
curl -X POST http://localhost:8000/flows/af9edd65-6393-58e2-9ae5-d5f012e714f4/run \
  -H "Content-Type: application/json" \
  -H "x-api-key: your-secret-key" \
  -d '{"input_value": "Hello, world!"}'

lfx run - Run flows directly

Execute a Langflow workflow and get results immediately.

uv run lfx run my_flow.json "What is AI?"

Options:

• --format, -f: Output format (json, text, message, result) (default: json)
• --verbose: Show diagnostic output
• --input-value: Input value to pass to the graph (alternative to positional argument)
• --flow-json: Inline JSON flow content as a string
• --stdin: Read JSON flow from stdin
• --check-variables/--no-check-variables: Check global variables for environment compatibility (default: check)

Examples:

# Basic execution
uv run lfx run simple_chat.json "Tell me a joke"

# JSON output (default)
uv run lfx run simple_chat.json "input text" --format json

# Text output only
uv run lfx run simple_chat.json "Hello" --format text

# Using --input-value flag
uv run lfx run simple_chat.json --input-value "Hello world"

# From stdin (requires --input-value for input)
echo '{"data": {"nodes": [...], "edges": [...]}}' | uv run lfx run --stdin --input-value "Your message"

# Inline JSON
uv run lfx run --flow-json '{"data": {"nodes": [...], "edges": [...]}}' --input-value "Test"

Complete Agent Example

Here's a step-by-step example of creating and running an agent workflow with dependencies:

Step 1: Create the agent script

Create a file called simple_agent.py:

"""A simple agent flow example for Langflow.

This script demonstrates how to set up a conversational agent using Langflow's
Agent component with web search capabilities.

Features:
- Uses the new flattened component access (cp.AgentComponent instead of deep imports)
- Configures logging to 'langflow.log' at INFO level
- Creates an agent with OpenAI GPT model
- Provides web search tools via URLComponent
- Connects ChatInput → Agent → ChatOutput
- Uses async get_graph() function for proper async handling

Usage:
    uv run lfx run simple_agent.py "How are you?"
"""

import os
from pathlib import Path

# Using the new flattened component access
from lfx import components as cp
from lfx.graph import Graph
from lfx.log.logger import LogConfig


async def get_graph() -> Graph:
    """Create and return the graph with async component initialization.

    This function properly handles async component initialization without
    blocking the module loading process. The script loader will detect this
    async function and handle it appropriately.

    Returns:
        Graph: The configured graph with ChatInput → Agent → ChatOutput flow
    """
    log_config = LogConfig(
        log_level="INFO",
        log_file=Path("langflow.log"),
    )

    # Showcase the new flattened component access - no need for deep imports!
    chat_input = cp.ChatInput()
    agent = cp.AgentComponent()

    # Use URLComponent for web search capabilities
    url_component = cp.URLComponent()
    tools = await url_component.to_toolkit()

    agent.set(
        model_name="gpt-4.1-mini",
        agent_llm="OpenAI",
        api_key=os.getenv("OPENAI_API_KEY"),
        input_value=chat_input.message_response,
        tools=tools,
    )
    chat_output = cp.ChatOutput().set(input_value=agent.message_response)

    return Graph(chat_input, chat_output, log_config=log_config)

Step 2: Install dependencies

# Install lfx (if not already installed)
uv pip install lfx

# Install additional dependencies required for the agent
uv pip install 'langchain-core>=0.3.0,<1.0.0' \
               'langchain-openai>=0.3.0,<1.0.0' \
               'langchain-community>=0.3.0,<1.0.0' \
               beautifulsoup4 lxml

Step 3: Set up environment

# Set your OpenAI API key
export OPENAI_API_KEY=your-openai-api-key-here

Step 4: Run the agent

# Run with verbose output to see detailed execution
uv run lfx run simple_agent.py "How are you?" --verbose

# Run with different questions
uv run lfx run simple_agent.py "What's the weather like today?"
uv run lfx run simple_agent.py "Search for the latest news about AI"

This creates an intelligent agent that can:

• Answer questions using the GPT model
• Search the web for current information
• Process and respond to natural language queries

The --verbose flag shows detailed execution information including timing and component details.

Input Sources

Both commands support multiple input sources:

• File path: uv run lfx serve my_flow.json
• Inline JSON: uv run lfx serve --flow-json '{"data": {"nodes": [...], "edges": [...]}}'
• Stdin: uv run lfx serve --stdin

Development

# Install development dependencies
make dev

# Run tests
make test

# Format code
make format

License

MIT License. See LICENSE for details.