Marimo Batch

What it does

The marimo-batch skill enables you to execute multiple Marimo notebooks sequentially within an AI agent workflow. This allows for complex, multi-step data processing and analysis where the output of one notebook serves as input for another. It facilitates automated pipelines and reduces manual intervention in repetitive tasks.

When to use it

• Automating a series of data cleaning and transformation steps using multiple Marimo notebooks.
• Building a pipeline that generates visualizations from data processed by several sequential notebooks.
• Running different stages of a machine learning project, such as training, evaluation, and deployment, each encapsulated in a separate notebook.
• Creating reproducible research workflows involving multiple analysis or modeling notebooks.

Key capabilities

• Sequential execution of Marimo notebooks.
• Passing output from one notebook to the next.
• Automated workflow creation using a list of notebook paths.

Example prompts

• "Run notebook1.marn then notebook2.marn."
• "Execute the following notebooks in order: /path/to/notebook_a.marn, /path/to/notebook_b.marn, /path/to/notebook_c.marn."
• "Process data using data_prep.marn and then visualize it with visualization.marn."

Tips & gotchas

Ensure that the notebooks you are batching are designed to accept input from previous steps, or define clear input mechanisms for each notebook within your workflow. Dependencies between notebooks must be clearly defined for proper execution order.