Jupyter Notebook Workflow Assistant
Create, refactor, and validate reproducible Jupyter notebooks with templates and a scaffold script.
The script reads templates and writes to a user-specified output path, refuses overwrite by default, and performs no network, credential, or external-service operations, providing some least-privilege and recovery protection. It does not document notebook data flows, sensitive-data handling, dependency security, or complete source attribution, so points are deducted.
The script uses the standard library and includes argument validation, template-existence checks, JSON-shape checks, and overwrite protection; the main path is plausible from static inspection. However, the supplied evidence omits the templates and contains no test suite or CI, so full installation and runtime behavior cannot be confirmed; the static calibration therefore limits the score.
The skill clearly covers experiments, tutorials, creation, conversion, and refactoring, with an explicit experiment/tutorial decision tree. Non-fit boundaries, exclusion triggers, Chinese-language support, and mainland-China environment fit are not specified; uv/Jupyter availability may affect some execution scenarios, so points are deducted.
Documentation is layered across triggers, workflow, references, script usage, and output conventions, with reasonably stable names and parameters. License metadata is unknown, versioning, changelog, maintenance ownership, and update path are unclear, and the repository README marks the repository deprecated, so points are deducted.
The intended output is a structured, reusable ipynb; the helper can create a titled template and reduce manual JSON errors, making the core scaffolding value clear. The templates are not included in the supplied evidence, content completion still requires manual work, and no representative output or execution result is verifiable, so the static effectiveness cap and uncertainty reduce the score.
The script and operating rules are auditable, and failure conditions and output behavior are visible in source. There is no test suite, CI, third-party execution evidence, or corroborating source, so the conclusion rests on limited static evidence and does not receive full marks.
- This is a static review only; the script and notebooks were not executed. Templates, icons, and license files are absent from the supplied evidence, so package completeness cannot be confirmed.
- The repository README says the repository is deprecated; verify the current maintenance channel and compatible Codex version before use.
- The skill may create or edit notebooks containing sensitive data, but provides no guidance for redaction, execution isolation, or output cleanup.
What it does & when to use it
This skill supports experiments, exploratory analysis, and teaching-oriented tutorials in `.ipynb` format. It selects an experiment or tutorial workflow based on the request and uses bundled templates plus `new_notebook.py` to create a clean starting notebook. It also guides targeted refactoring of existing notebooks, focused runnable cells, concise explanations, and top-to-bottom validation. The repository is deprecated; the README directs users to the OpenAI Plugins repository for current skill and plugin examples.
Identifies whether a notebook request is an experiment, tutorial, or refactor; runs $CODEX_HOME/skills/jupyter-notebook/scripts/new_notebook.py; loads either assets/experiment-template.ipynb or assets/tutorial-template.ipynb; updates the title and writes the notebook to a chosen output path; structures focused code cells and explanatory Markdown; and, when possible, executes the notebook top to bottom and applies the quality checklist.
- A data analyst starting a hypothesis-driven exploration can use the experiment workflow for a consistent notebook structure.
- A researcher comparing prompt variants can scaffold a reproducible experiment notebook quickly.
- A teaching author creating a step-by-step, audience-specific walkthrough can use the tutorial workflow to organize code and explanations.
- A developer converting rough scripts or notes into a readable notebook can follow the refactoring guidance while preserving intent.
Pros & cons
- Provides distinct workflows for experiments and tutorials.
- Templates and a standard-library scaffold script reduce hand-authored notebook JSON errors.
- Includes guidance for refactoring existing notebooks and validating execution.
- Defines temporary and final output conventions for cleaner projects.
- The repository README explicitly marks the repository as deprecated.
- The source provides no test suite, platform matrix, or execution results.
- Local notebook execution may require installing `jupyterlab` and `ipykernel`.
- The skill covers notebook workflow rather than domain-specific analysis, data access, or model capabilities.
How to install
The skill is located at skills/.curated/jupyter-notebook/. The README states that curated skills can be installed in Codex with: $skill-installer jupyter-notebook. Restart Codex after installation so it can load the skill. The source does not state the license; it says to inspect LICENSE.txt inside the individual skill directory, but its contents were not provided.
How to use
First define the objective, audience, and completion criteria, then choose experiment or tutorial. Example trigger: “Create a tutorial notebook titled Intro to embeddings.” To scaffold it, run: uv run --python 3.12 python "$JUPYTER_NOTEBOOK_CLI" --kind tutorial --title "Intro to embeddings" --out output/jupyter-notebook/intro-to-embeddings.ipynb. $JUPYTER_NOTEBOOK_CLI defaults to $CODEX_HOME/skills/jupyter-notebook/scripts/new_notebook.py.