notebook-ai-agents-skill

What problem does it solve?

Building reproducible, narrative-first notebooks is hard when environments drift, code runs are non-deterministic, and sharing results lacks an end-to-end verification gate. This skill provides a structured approach to authoring notebooks that load data reproducibly, use per-directory Pixi environments, and validate run-all execution.

Core Features & Use Cases

• Narrative-first notebook structure with Markdown-guided code cells.
• Per-directory Pixi environments to ensure reproducible kernels.
• DuckDB-backed data loading patterns with project-relative paths.
• End-to-end validation through scripts/execute_notebook.py for deterministic outcomes.
• Guidance for plotting and reporting in a consistent style.

Quick Start

Run a full end-to-end validation on a notebook using the included execute_notebook script.