• Example Repository for Custom Python Component
  • ⚙️ Configuration
  • 📦 Managing dependencies
  • 🚀 Running locally…
    • 🐋 …in Docker
    • 💜 …in virtual environment
    • 💔 …locally without uv

CrewAI Boilerplate for Keboola Custom Python Component

This repository provides a boilerplate for running CrewAI flows inside Keboola's Custom Python Component. Use this as a starting point for integrating CrewAI automations into your Keboola projects.

The structure and configuration are designed to make it easy to:

• Develop and test CrewAI flows locally
• Seamlessly deploy and run them in Keboola's custom python component environment

⚙️ Configuration

Before running the project, you must create a configuration file:

1. Copy the example config file:

   cp data/config.json.dist data/config.json

2. Edit data/config.json and provide your OpenAI API key in the parameters section (as #OPENAI_API_KEY).

📦 Managing dependencies

Dependencies are managed using uv. The keboola.component package is listed as a dev dependency, as it is already included in the Keboola environment. To add dependencies:

• Use uv add for runtime dependencies (these will be installed when running the project in Keboola).
• Use uv add --dev for development dependencies (these won't be installed in Keboola).

🚀 Running locally…

🐋 …in Docker

When you have Docker installed, the easiest way to run this example is using docker compose:

docker compose build  # voluntary when running for the 1st time
docker compose run dev

This way, the configuration and source files are always updated during the build.