FNGEN: Generative infrastructure for Python functions

fngen ("f-n-gen") is the fastest way for Python developers to deploy powerful applications. We provide generative infrastructure for Python functions, turning your code into production-ready solutions in minutes.

The fngen Method: Functions + Powers = Infrastructure.

  1. Write your Python function.
  2. Assign it a Power (e.g., @webapp, @task_worker) to define its role.
  3. Run fngen push.

fngen generates the necessary Functional Infra – servers, load balancers, task queues, HTTPS, and deployment pipelines – tailored to your function's Power. No manual setup, no complex configs.

Key Benefits:

Quickstart

1. Download fngen

pip install fngen

2. Get your API key

Download API Key

3. Push a deployment package

Check out our repo of cool examples to get inspired + get started!

Hello world

Let’s say you want to deploy a simple FastAPI web app.

Here’s the vanilla version:

# my_website.py
from fastapi import FastAPI
from fastapi.responses import HTMLResponse

app = FastAPI()

@app.get('/')
def my_fastapi_page():
    return HTMLResponse('<h1>FastAPI on fngen</h1>')

Now, here’s how you empower it with fngen:

# my_website.py
from fastapi import FastAPI
from fastapi.responses import HTMLResponse
from fngen.powers import webapp

app = FastAPI()

@app.get('/')
def my_fastapi_page():
    return HTMLResponse('<h1>FastAPI on fngen</h1>')

# fngen-specific addition
@webapp(framework='fastapi', compute='server')
def live_app():
    return app

Then:

fngen push my_website.py

💥 Boom — fngen provisions infrastructure, deploys your app, handles HTTPS, and gives you a public URL.


Table of Contents


Powers

fngen activates your Python functions using Powers. Each Power tells fngen what to do with your function — deploy it as a web app, run it on a schedule, generate a UI for it, and more.


@webapp(framework='...', compute='...')


FastAPI Example

# fastapi_app.py
from fastapi import FastAPI
from fngen.powers import webapp

app = FastAPI()

@app.get("/ping")
def ping():
    return {"ping": "pong"}

@webapp(framework='fastapi', compute='server')
def serve_fastapi():
    return app

# Run: fngen push fastapi_app.py

Flask Example

# flask_app.py
from flask import Flask
from fngen.powers import webapp

app = Flask(__name__)

@app.route("/")
def hello():
    return "Hello from Flask on fngen!"

@webapp(framework='flask', compute='server')
def serve_flask():
    return app

# Run: fngen push flask_app.py

Django Example

# myproject/asgi.py
import os
from django.core.asgi import get_asgi_application
from fngen.powers import webapp

os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'myproject.settings')
application = get_asgi_application()

@webapp(framework='django', compute='server')
def serve_django():
    return application

# Run: fngen push myproject/asgi.py

@task_worker(compute='...')

Purpose: Deploys a function to process background jobs or tasks asynchronously.

Compute options:

How it Works: fngen provisions the necessary infrastructure to receive and execute jobs. You can dispatch tasks to the worker by sending payloads, and the function processes them in the background.

Function Signature: The function can accept any number of parameters with any Python type annotations (str, int, dict, float, custom classes, etc.). fngen will smartly route incoming task data to your function, matching types and handling deserialization.

Example:

from fngen.powers import task_worker

@task_worker(compute='server')
def process_order(order_id: int, customer_email: str, items: list[dict]):
    """
    Processes an order by ID, customer email, and a list of purchased items.
    """
    print(f"Processing order {order_id} for {customer_email}")
    print(f"Items: {items}")

Dispatch a task via the fngen SDK:

from fngen import platform

platform.run_task('your_project', 'hello_world', {'name': 'Aang'})

or via the fngen CLI:

fngen run_task your_project hello_world '{"name": "Aang"}'

Config Overview + Roadmap

@webapp Configuration Options

Deploys your Python web application.

Option Value Status Description
framework 'fastapi' Supported Deploy an application built with FastAPI.
framework 'flask' Coming Soon Deploy an application built with Flask.
framework 'django' Coming Soon Deploy an application built with Django (ASGI/WSGI).
compute 'server' Supported Deploys to always-on server instance(s) defined in fngen.yml.
compute 'serverless' Coming Soon Deploys to event-driven infrastructure that scales to zero when idle.

@task_worker Configuration Options

Deploys a function to process background jobs or tasks asynchronously.

Option Value Status Description
compute 'server' Supported Deploys to always-on worker server instance(s) defined in fngen.yml.
compute 'serverless' Coming Soon Deploys to event-driven worker infrastructure that scales to zero when idle.

Projects

Your deployments, code, packages, etc. are organized into Projects.

You can create a project with the following command:

fngen project create [project_name]

List all projects in your account:

fngen project list

Defining Your Deployment

When you run fngen push, fngen packages your code and relevant configuration files. These files help define dependencies, infrastructure resources, and exactly which files are included or excluded.

Code structure and packaging

When you run fngen push [path/to/code_dir], the specified directory becomes your deployment package. fngen will package and upload the contents of this directory to the platform for building and deployment.

By default, fngen includes all files and subdirectories within the package directory, with specific rules for inclusion and exclusion defined by .gitignore, .fnignore, and .fninclude files (explained in detail below).

During the deployment process, fngen scans all .py files within your package to discover functions decorated with fngen.powers decorators.

The essential files for a basic deployment are your Python code and a fngen.yml file defining your project resources. Note: fngen.yml must be located at the root of your deployment package directory.

Below are some common example directory structures for a deployment package.

Simple File Structure

Here's a minimal structure with a single Python file containing your code, and a fngen.yml file defining your resources.

.
├── hello.py
└── fngen.yml

Structure with requirements.txt

In most cases, you'll need third-party Python packages. Simply include a requirements.txt file at the root of your package, and fngen will automatically install the dependencies during the deployment build process.

.
├── requirements.txt
├── hello.py
└── fngen.yml

Structure with multiple modules

You are not confined to a single Python module. You can organize your code across as many .py files as needed. fngen will scan all of them for decorated functions.

.
├── requirements.txt
├── hello.py
├── llm.py
├── run_analysis.py
└── fngen.yml

Structure with an inner Python package

It's common practice to organize larger codebases into Python packages using subdirectories with an __init__.py file. This standard structure is fully supported.

.
├── my_project
│   ├── __init__.py
│   ├── annotate_video.py
│   └── recommend_videos.py
├── requirements.txt
└── fngen.yml

Structure including git-ignored assets

It's good practice to exclude large 'assets' (like data files, videos, images) from your version control (Git) repository using .gitignore. However, you will often want to include these in your deployment package. You can achieve this by adding entries to a .fninclude file at the root of your package. fngen respects .gitignore for exclusions but prioritizes .fninclude for inclusions. You can also use .fnignore for explicit exclusions (see .fnignore and .fninclude sections for details).

.
├── .gitignore
├── .fninclude
├── assets
│   ├── big_video.mp4
│   └── gorilla.png
├── my_project
│   ├── __init__.py
│   ├── annotate_video.py
│   └── recommend_videos.py
├── requirements.txt
└── fngen.yml

requirements.txt

If your code includes a requirements.txt file, fngen will automatically install dependencies for each run mode.

Under the hood, it simply runs:

pip install -r requirements.txt

before setting up the deployment or runtime environment.

You don't need to do anything special — just make sure your dependencies are correctly listed in your requirements.txt.

Best practice: Always commit your requirements.txt alongside your code.


fngen.yml

Your fngen.yml file, placed at the root of your project directory, is where you define project-level configuration and the infrastructure Resources required by your fngen Powers.

Currently, fngen.yml is primarily used to define configurations for server compute resources. Support for configuring serverless resources will be added in the future.

Defining Your Server Fleet

Powers that require persistent, always-on compute, such as @webapp(compute='server') or @task_worker(compute='server'), rely on server resources defined in your fngen.yml under the servers: key.

You define the composition of your server fleet by listing one or more server configurations using name, size, and region. fngen will provision a fleet of server instances based on these definitions. For detailed pricing information on server sizes, please refer to the Pricing section.

Example:

# fngen.yml
servers:
  - id: web-server-class # Required: A unique name for this server definition within your project. Used for identification in the config.
    size: s              # Required: The server type tier. Choose one from the table below (see Pricing section).
    region: us-west      # Required: The geographic region for deployment (e.g., us-west, us-east, eu-central).
# You can define multiple server types to include in your fleet:
  - id: worker-server-class
    size: xl
    region: us-west

Explanation: In this example, you are defining a server fleet that will include server instances of size s located in us-west and server instances of size xl also in us-west. The name field (web-server-class, worker-server-class) is purely for identifying these different definitions within your fngen.yml.

How compute='server' Powers Use Your Fleet

When you deploy your project with fngen push, any function decorated with a Power using compute='server' (like @webapp or @task_worker) will be deployed to every single server instance within the fleet you defined in fngen.yml.

By defining multiple server configurations (e.g., one s and one xl), you create a heterogeneous fleet, and your compute='server' workloads will run on instances of both types.

(Future: This file may include configurations for serverless compute or other fngen-managed resources.)

.fninclude

Use .fninclude to include .gitignore'd files in your deployment package when you run fngen push.

Typically, you don't want to include certain resources in your git repo, but you do want to include them in your deployment package.

.fnignore

Use .fnignore to ignore files from being included in your deployment package when you run fngen push.

Uses .gitignore syntax.

Note: fngen push also respects your .gitignore; .fnignore is used to ignore stuff that you may want to include in your git repo. (This is rare, but it can happen).

Environment Variables

Managing application configuration, secrets, and credentials securely is essential. fngen provides a built-in mechanism for managing environment variables for your projects.

Purpose

Environment variables are key-value pairs injected into the runtime environment where your functions are executed. They are commonly used for:

Setting Environment Variables

You can upload environment variables to your fngen project using the fngen project set_env command. This command reads variables from a standard .env file.

  1. Create a .env file: In your project directory (or any location), create a file named .env (or any name you prefer). Add your variables using the KEY=VALUE format, one per line:

    DATABASE_URL=postgres://user:password@host:port/database
    STRIPE_SECRET_KEY=sk_live_...
    API_TIMEOUT_SECONDS=30
    

    Important: Do not commit your .env file containing secrets to your version control system (like Git). Add it to your .gitignore file.

  2. Upload with the CLI: Use the set_env command, specifying your project name and the path to your .env file.

    fngen project set_env my-project .env
    

This command securely uploads the variables to the fngen platform. They are encrypted at rest and injected into the runtime environment of your project's deployed functions. Uploading a variable with the same key will overwrite its existing value.

Accessing Environment Variables in Your Code

Inside any function deployed with a fngen Power, you can access environment variables using Python's standard os module:

import os

from fngen.powers import webapp

@webapp(framework='fastapi', compute='server')
def my_app_with_env():
    from fastapi import FastAPI
    app = FastAPI()

    @app.get("/secret")
    def get_secret():
        # Accessing an environment variable
        secret_key = os.environ.get("MY_SUPER_SECRET")
        if not secret_key:
            return {"error": "MY_SUPER_SECRET not set"}, 500
        return {"secret": secret_key}

    return app

# You would set MY_SUPER_SECRET using `fngen project set_env my-project .env`
# where .env contains: MY_SUPER_SECRET=its_a_secret!

Environment variables are available to all functions within a project, regardless of the Power used.


Authentication and API Keys

To interact with the fngen platform and deploy your code, you need to authenticate your fngen CLI and SDK using an API key. Your API key links your local tools to your fngen account and projects.

Obtaining Your API Key

Your API key is generated within your fngen account dashboard (details on accessing the dashboard will be provided separately). Treat your API key like a password – keep it secure and do not share it publicly.

Configuring Authentication

You can configure your API key for the fngen CLI and SDK in two primary ways:

  1. Configuration File (Recommended):

    • Create a directory named .fngen in your user's home directory (~/.fngen/).

    • Inside this directory, create a file named credentials.yml.

    • Add your API key to this file in the following format:

      api_key: your_fngen_api_key_here
      
    • Make sure this file has restricted permissions (chmod 600 ~/.fngen/credentials.yml) to protect your API key.

  2. Environment Variable:

    • Set the FNGEN_API_KEY environment variable in your terminal session or shell profile.

      export FNGEN_API_KEY=your_fngen_api_key_here
      
    • This method is often used in CI/CD pipelines or temporary setups. Note that environment variables can sometimes be less secure than a properly permissioned configuration file.

The fngen CLI and SDK will check for the environment variable first, and if not found, will look for the credentials.yml file.

Verifying Your Authentication

Use the fngen connect command to verify that your authentication is set up correctly and you can connect to the fngen platform.

fngen connect

If successful, this command will confirm your connection and display the email address associated with your account.


Logging and Monitoring

Observability is key to understanding the behavior of your deployed applications and tasks, diagnosing issues, and monitoring performance. fngen provides integrated logging capabilities for all your deployed functions.

What Gets Logged?

fngen captures standard output (stdout) and standard error (stderr) from your Python functions. This means any print() statements or output from standard Python logging libraries (like logging) within your decorated functions will be automatically collected and made available through the fngen platform.

Additionally, fngen platform events related to your project (deployments, task dispatches, scaling events, errors) are also included in the logging stream.

Accessing Logs via the CLI

The primary way to access logs is using the fngen logs command.

fngen logs [project_name] [options]

As detailed in the CLI section, this command allows you to view recent logs, follow logs in real-time, and filter by Power, function, task ID, time range, or log level.

Examples:

Using Python's logging Module

While print() works fine for simple output, using Python's built-in logging module provides more structure, severity levels, and context for your logs. fngen fully supports this.

import logging
from fngen.powers import task_worker

# Configure basic logging (optional, but good practice)
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

@task_worker(compute='server')
def process_data(data_payload: dict):
    """
    Processes data received via a task.
    """
    try:
        logger.info(f"Received data payload: {data_payload}")
        # ... processing logic ...
        logger.info("Data processing successful")
        # You can log at different levels:
        # logger.warning("Something unusual happened")
        # logger.error("An error occurred during processing")
    except Exception as e:
        logger.exception(f"Error processing data: {e}") # Logs error and traceback
        raise # Re-raise the exception if needed

Logs generated using logging.info(), logging.error(), etc., will appear in the fngen logs output with their respective severity levels.

Future Monitoring Features

While basic logging is available now, future versions of fngen will include more advanced monitoring features such as:


fngen CLI

The fngen Command Line Interface (CLI) is your primary tool for interacting with the fngen platform, managing your projects, and deploying your code.

Ensure you have the fngen CLI installed and configured with your API key before proceeding.

pip install fngen

Authentication

The CLI requires authentication to interact with your fngen account. You can configure your credentials in ~/.fngen/credentials.yml or by setting the FNGEN_API_KEY environment variable.

Use the connect command to verify your setup.

Commands

Here are the most commonly used fngen CLI commands:

fngen connect

Verify your authentication credentials and connectivity to the fngen platform.

fngen connect

Checks if your API key (from ~/.fngen/credentials.yml or FNGEN_API_KEY) is valid. If successful, it will return the email address associated with the authenticated account.

fngen push [project_name] [path/to/package]

Package and deploy your Python code to a specific project.

fngen push [project_name] [path/to/package]

This command is the core deployment action.

fngen logs [project_name]

View logs generated by your deployed functions and the fngen platform for your project.

fngen logs [project_name] [options]

Accesses the centralized logging stream for your project. Useful for debugging, monitoring, and seeing the output of your print() statements or logging calls. If [project_name] is omitted when run inside a project directory, it defaults to the current project.

fngen project list

List all fngen projects associated with your account.

fngen project list

Displays a list of your project names.

fngen project create [project_name]

Create a new fngen project in your account.

fngen project create [project_name]

Project names must be unique across your fngen account. Creating a project sets up the necessary infrastructure scaffolding on the platform side.

fngen project set_env [project_name] [path/to/.env]

Securely upload environment variables from a .env file to your project.

fngen project set_env [project_name] [path/to/.env_file]

Reads key-value pairs from the specified .env file (standard format: KEY=VALUE). These variables are securely stored on the platform and automatically injected into the runtime environment whenever your project's functions are executed (web requests, tasks, etc.). Existing variables with the same name will be overwritten.

fngen project destroy [project_name]

Permanently delete a project and all associated resources (servers, data, configurations, logs).

fngen project destroy [project_name]

This is a destructive and irreversible action. You will typically be prompted to confirm before deletion proceeds. Use with caution.

(Future: Mention fngen --help and fngen [command] --help for more detailed CLI help).


Pricing

fngen pricing is based on the server resources you provision and utilize for your applications and tasks. We aim for simple, predictable costs.

Server Pricing

When you use compute='server' for your webapp or task_worker Powers, you define a fleet of servers in your fngen.yml file based on size and region. You are charged based on the size and duration of these server instances.

The table below shows the available server types, their specifications, and estimated monthly pricing per instance.

Server Type (size) vCPU RAM Disk Transfer Est. Cost (USD/mo)
xs 1 0.5 GB 10 GB 0.5 TB 8
s 1 1 GB 25 GB 1 TB 12
m 1 2 GB 50 GB 2 TB 24
l 2 2 GB 60 GB 3 TB 36
xl 2 4 GB 80 GB 4 TB 48
2xl 4 8 GB 160 GB 5 TB 96
3xl 8 16 GB 320 GB 6 TB 192

(Pricing for serverless compute and other future resource types will be added here.)

What's Included / No Extra Cost

fngen abstracts away significant infrastructure complexity and management overhead. The following are included as part of the platform fee covered by your server instance costs:

You only pay for the server instances you define in your fngen.yml or the usage of future metered resources like serverless functions.

Pricing alt:

Hobby Plan

$0 / Month Subscription

Usage-based pricing for infra resources (servers, serverless_clusters, websocket_backbones)

Boosted Plan

$20 / Month Subscription

Access to @ai_* powers

Enterprise Plan

Contact for pricing. On-prem installation available.


FAQ

Why Python?

There's a great saying that goes:

"python isn't the best language for anything, but it's the second best language for everything"

So, when building a platform focused on delivering powerful, high-leverage solutions with code, it seems like the best all-purpose language to focus on.

Plus, we just love Python.

P.S., Need other tools? You can often integrate them via a standard Python interface within your fngen-activated functions.


Why Python functions specifically?

Python functions serve as a versatile and well-defined unit for specifying discrete pieces of logic. By centering fngen around functions, we provide a simple, consistent contract for activating various Powers. This approach promotes modularity, testability, and allows fngen to generate the appropriate execution environment and infrastructure tailored precisely to that function's defined role.


Functional Infra: The Power of Purpose-Built Infrastructure

Building and deploying powerful applications requires more than just writing code. It requires supporting infrastructure – things like servers, networks, load balancers, message queues, and more. Traditionally, setting up and managing this infrastructure has been a complex and time-consuming task for developers.

fngen introduces the concept of Functional Infra. This refers to the minimum, pre-packaged infrastructure automatically provisioned and managed by fngen specifically to support a certain Python function and its designated Power.

Think of each Power as a blueprint for a specific type of application function, and Functional Infra as the engine and support system needed to make that blueprint a reality in a production environment.

Here are some examples of how Functional Infra pairs with fngen Powers:

The Value of Functional Infra:

By providing Functional Infra, fngen significantly reduces the operational burden on developers. You don't need to become an infrastructure expert to deploy a web app or set up a task queue. You simply define what your function is meant to do using a Power, and fngen handles provisioning, configuring, and managing the underlying infrastructure required for that purpose.

This allows you to:

As fngen grows, we will continue to introduce new Powers, each bringing its own flavor of purpose-built Functional Infra to support a wider range of Python use cases – all while maintaining the simple, function-centric "annotate and push" workflow.


TL;DR

  1. Write a Python function.
  2. Give it a Power (decorate it).
  3. fngen push it.

fngen handles the rest.


Roadmap

This section outlines the current capabilities of fngen and provides a look ahead at planned features and enhancements.

General Roadmap & Future Enhancements

The following areas represent ongoing development and future additions to the fngen platform and documentation:

Planned Powers (Coming Soon)

These Powers are planned for future releases:

We are actively working on expanding the capabilities of fngen to cover a wider range of Python application patterns. Stay tuned for updates!