Anedya Sparks Runtime SDK

Build Python functions for Anedya Sparks with this SDK. You write a handler for each invocation; the SDK connects to the Sparks runtime, delivers events, and returns your handler's result.

Project structure

Create a project with packages/ for the function's installed dependencies and src/ for your function. Use a separate local virtual environment for development. Sparks runs src/main.py as the entrypoint.

my-spark-function/
├── packages/        # Function dependencies included in the asset
├── src/
│   └── main.py      # Required entrypoint
└── .venv/           # Local development environment; do not upload

Set up the environment from your project directory:

Use Python 3.13 or newer.

python -m venv .venv
source .venv/bin/activate
mkdir -p packages src
python -m pip install --target packages anedya-sparks-runtime

On Windows, activate the environment with .venv\Scripts\activate.

Write a function

Put this in src/main.py:

import json

import anedya.sparks_runtime as sparks


@sparks.handler
def handle(event: sparks.Event) -> str:
    data = event.json_payload()
    result = {"message": f"Hello, {data.get('name', 'Sparks')}!"}
    return json.dumps(result)


sparks.run()

The handler receives one sparks.Event. Use event.json_payload() for a JSON payload, or inspect event.request_id, event.trigger_type, and other event fields. Return bytes, str, or None; strings are encoded by the SDK. Keep sparks.run() at the end of main.py: it starts the invocation loop and blocks while the function serves requests.

Package for upload

The setup command installs the SDK and its dependencies into packages/. Install additional function dependencies there with python -m pip install --target packages <package>. From the project directory, zip packages/ and src/; the local .venv/ is only for development and does not need to be included.

zip -r function.zip packages src

Upload function.zip as the function asset.