PyAirbyte

PyAirbyte brings the power of Airbyte to every Python developer. PyAirbyte provides a set of utilities to use Airbyte connectors in Python.

Getting Started

Watch this Getting Started Loom video or run one of our Quickstart tutorials below to see how you can use PyAirbyte in your python code.

• Basic Demo
• CoinAPI
• GA4
• Shopify
• GitHub
• Postgres (cache)

Airbyte MCP

The Airbyte MCP is a Model Context Protocol server that connects AI agents to your Airbyte organization. Your agent works with the same permissions you have in Airbyte. Use it with clients such as Claude, ChatGPT, and VS Code Copilot.

With the Airbyte MCP, your agent can:

• Build and manage pipelines: deploy sources and destinations, create connections, select streams, and start syncs.
• Troubleshoot pipelines: check sync status, read job logs, and inspect connector configurations.
• Read data directly: get live records from supported sources and query supported destinations without waiting for a sync.
• Learn connector capabilities through connector skills, which describe supported entities, actions, and parameters.
• Find answers in Airbyte's documentation and other Airbyte knowledge sources.

Direct reads are read-only and only work with supported connectors after an organization admin enables the context layer. They do not work with connectors whose secrets are stored in an external secrets manager.

The hosted server works with all editions of Airbyte Cloud and Airbyte Flex. You can run it locally with self-managed Airbyte Core, but local deployments have known issues.

To connect, use https://mcp.airbyte.com/mcp and sign in with your Airbyte Cloud account through OAuth. See the installation instructions and complete tool list.

Connector Installation

Declarative Source Installation

For Declarative Sources defined in YAML, the installation process will is to simply download the yaml file from the connectors.airbyte.com public URLs, and to run them directly as YAML.

Declarative sources have the fastest download times, due to the simplicity and each of install.

In some cases, you may get better stability by using docker_image=True in get_source()/get_destination(), due to the fact that all dependencies are locked within the docker image.

Python Installation

Generally, when Python-based installation is possible, it will be performed automatically given a Python-based connector name.

In some cases, you may get better stability by using docker_image=True in get_source()/get_destination(), due to the fact that all dependencies are locked within the docker image.

Installing Connectors with uv

By default, beginning with version 0.29.0, PyAirbyte defaults to uv instead of pip for Python connector installation. Compared with pip, uv is much faster. It also provides the unique ability of specifying different versions of Python than PyAirbyte is using, and even Python versions which are not already pre-installed on the local workstation.

If you prefer to fall back to the prior pip-based installation methods, set the env var AIRBYTE_NO_UV=true.

Installing Connectors With a Custom Python Version

In both get_source() and get_destination(), you can provide a use_python input arg that is equal to the desired version of Python that you with to use for the given connector. This can be helpful if an older connector doesn't support the version of Python that you are using for PyAirbyte itself.

PyAirbyte itself requires Python 3.11 or 3.12, but connectors can still be installed and run under older Python versions. For example, assuming PyAirbyte is running on Python 3.12, you can install a connector using Python 3.10.17 with the following code snippet:

import airbyte as ab

source = ab.get_source(
    "source-faker",
    use_python="3.10.17",
)

Installing Connectors with Docker

For any connector (get_source()/get_destination()), you can specify the docker_image argument to True to prefer Docker over other default installation methods or docker_image=MY_IMAGE to leverage a specific docker image tag for the execution.

Contributing

To learn how you can contribute to PyAirbyte, please see our PyAirbyte Contributors Guide.

Frequently asked Questions

1. Does PyAirbyte replace Airbyte? No. PyAirbyte is a Python library that allows you to use Airbyte connectors in Python, but it does not have orchestration or scheduling capabilities, nor does is provide logging, alerting, or other features for managing pipelines in production. Airbyte is a full-fledged data integration platform that provides connectors, orchestration, and scheduling capabilities.

2. What is the PyAirbyte cache? Is it a destination? Yes and no. You can think of it as a built-in destination implementation, but we avoid the word "destination" in our docs to prevent confusion with our certified destinations list here.

3. Does PyAirbyte work with data orchestration frameworks like Airflow, Dagster, and Snowpark, Yes, it should. Please give it a try and report any problems you see. Also, drop us a note if works for you!

4. Can I use PyAirbyte to develop or test when developing Airbyte sources? Yes, you can. PyAirbyte makes it easy to test connectors in Python, and you can use it to develop new local connectors as well as existing already-published ones.

5. Can I develop traditional ETL pipelines with PyAirbyte? Yes. Just pick the cache type matching the destination - like SnowflakeCache for landing data in Snowflake.

6. Can PyAirbyte import a connector from a local directory that has python project files, or does it have to be pip install Yes, PyAirbyte can use any local install that has a CLI - and will automatically find connectors by name if they are on PATH.

Changelog and Release Notes

For a version history and list of all changes, please see our GitHub Releases page.