probabl-aiprobabl-ai/skore-skills/skills/export-ml-notebook/SKILL.md

export-ml-notebook

Skill144 starsChanged 17 days agoBSD-3-ClauseView on GitHub

Write a jupytext percent file out as a notebook with empty outputs. `data_analysis/`, `experiments/`, and `audit/` are filled with empty outputs. Pass `--html` when that notebook should also become a viewer on the site. Kernel `notebook convert` stays for a source that is itself the execution. Trigger on notebook, ipynb, HTML, or executed-report requests, or when export-ml-project dispatches notebooks.

1 other project uses a version of this.

Adopted by

Most recently taken up by

probabl-ai/skillsword for word, first seen Oct 27, 2026

Other versions1 project uses a version of this.
Word for word · 1
probabl-ai/skills★ 144100% same
Edited · 0
None found.
Wrote their own version · 0
None found.
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository26 other files it gives its agents.
- AGENTS.mdAGENTS.md
- add-python-packageSkill
- audit-ml-pipelineSkill
- build-ml-pipelineSkill
- choose-python-librarySkill
- evaluate-ml-pipelineSkill
- explore-ml-dataSkill
- export-ml-projectSkill
- export-ml-siteSkill
- frame-ml-problemSkill
- manage-ml-backlogSkill
- model-ml-pipelineSkill
- persist-ml-gitSkill
- plot-ml-figureSkill
- research-ml-practiceSkill
- review-ml-choicesSkill
- review-ml-experimentSkill
- search-ml-literatureSkill
- setup-gitSkill
- setup-ml-projectSkill
- setup-python-envSkill
- setup-workspaceSkill
- shape-user-ideaSkill
- smoke-test-ml-pipelineSkill
- sync-ml-reportsSkill
- triage-ml-taskSkill

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 2 are distinctive, and 10 are very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
10
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Only policy.notebooks true enables automatic notebook materialization.
line 40
Very specific:
When true, exploratory analysis uses materialize.py as its
line 44
Very specific:
Run status and read policy.notebooks.
line 58
Very specific:
If policy.notebooks is null: persist python -m skore_skills policy set notebooks true.
line 74
Very specific:
If policy.notebooks is false: say notebooks are off.
line 78
Very specific:
If the user wants the notebook on the site, load add-python-package for nbconvert (agent) and pass --html (writes a self-contained …
line 95
Very specific:
After --html, if policy.site is true, export-ml-site is installed, run python -m skore_skills site build --if-stale so the viewer is …
line 118
Distinctive:
If that skill is missing, or the retry still fails, name the error in one line.
line 126
Distinctive:
Do not git commit or git end-turn.
line 136
Very specific:
Do not run cells run as a substitute for fill.
line 137
Very specific:
Do not pixi add / uv add; load add-python-package.
line 138
Very specific:
Do not name --html or python -m skore_skills site build unless the user asked for HTML or a site viewer.
line 140

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

SKILL.md

Export ML Notebook

Source of truth stays the # %% .py. This skill only writes derived .ipynb (and optional .nb.html). Do not rewrite the .py from the notebook. cells run is not a substitute.

Human-facing prose

Details: setup-workspace references/human_facing_prose.md. Tell the user the written .ipynb (and HTML viewer) paths. Do not quote notebook convert, --html, or site build as something they should run.

While policy.notebooks is true, explore-ml-data fills the analysis file it wrote. model-ml-pipeline and evaluate-ml-pipeline fill the experiment script and, when it exists, audit/<stem>.py, with empty outputs. audit-ml-pipeline fills audit/<stem>.py only on a direct close. This skill owns on-demand fills, other sources, and turning the flag on when it is null or false. A data-analysis, experiment, or audit source uses notebook fill, not notebook convert.

Workflow checkpoint execution

Stage owners read policy before executing a human # %% source. Only policy.notebooks true enables automatic notebook materialization; null and false keep the stage's non-notebook execution and do not change policy.

When true, exploratory analysis uses materialize.py as its one run, then notebook fill writes the .ipynb with empty outputs. Add --html only when policy.site is also true. An experiment or audit file is the same: notebook fill writes the .ipynb with empty outputs, and materialize.py is the run that writes the digest and the HTML viewers. Do not notebook convert those files and do not cells run them. Kernel notebook convert stays only for a source that is itself the execution (the unfitted experiment snapshot). A generated notebook records its source fingerprint; loop notebooks fills or converts again only when the source or required HTML changed.

Lookup

Run status and read policy.notebooks. Take the first row that matches. Fill leaves code-cell outputs empty. Do not stop after reading the table.

`policy.notebooks`	Request	Then
`null`	any	step 2, then match this table again. Do not take the convert row only because it is listed first.
`false`	any	step 3. Do not fill until the policy is `true`.
`true`	the user asked for HTML, a site viewer, or the notebook on the site	step 4 `notebook fill --html`, then step 5. `data_analysis/`, `experiments/<stem>.py`, and `audit/<stem>.py` stay `notebook fill`, not `notebook convert`.
`true`	the source is itself the execution (unfitted experiment snapshot, before `skore.evaluate`) and the user did not ask for HTML or a site viewer	kernel `notebook convert`, not `notebook fill`
`true`	notebook, ipynb, or convert, with no HTML and no site viewer	step 4 `notebook fill` only. No `--html`. No `site build`.

Sequence

python -m skore_skills status. Read policy.notebooks and skills.
If policy.notebooks is null: persist python -m skore_skills policy set notebooks true. Do not AskUserQuestion. Then load add-python-package for jupytext and nbformat (agent) and continue.
If policy.notebooks is false: say notebooks are off; offer to turn them on. Do not fill until the policy is true.
Fill the requested percent file (default data_analysis/data_analysis.py when the user did not name one):
```
python -m skore_skills notebook fill data_analysis/data_analysis.py
```
Fill does not execute the file and leaves code-cell outputs empty. It needs jupytext and nbformat, not nbclient, ipywidgets, or IPython. Do not start EDA, an evaluation, or an audit from this skill to produce those outputs. experiments/<stem>.py and audit/<stem>.py are filled the same way. Missing jupytext / nbformat / nbconvert stays a one-line skip.
If the user wants the notebook on the site, load add-python-package for nbconvert (agent) and pass --html (writes a self-contained <stem>.nb.html next to the .py). The site links that file from the associated exploratory data analysis or design report. It opens as a page of its own.
Optional --out path.ipynb. Keep *.ipynb gitignored unless the user asks setup-git to track them.
Fill-only requests (notebook / ipynb / convert, no HTML or site viewer): run only that notebook fill line, with no --html. Tell the user the written .ipynb path. Do not mention --html or site build as an optional aside.
Kernel notebook convert is only for a source that is itself the execution: the unfitted experiment snapshot, before skore.evaluate exists. Convert injects %matplotlib inline for that kernel run so seaborn / matplotlib last expressions emit image/png, then strips that setup cell from the written notebook. Do not put %matplotlib inline in the .py (style / ruff would reject it). If that convert fails because ipywidgets is missing, load add-python-package for it (agent) and convert again.
After --html, if policy.site is true, export-ml-site is installed, run python -m skore_skills site build --if-stale so the viewer is packaged. Run --html and site build only when the user asked for HTML or a site viewer. Fill without --html does not rebuild the site. Skip in one line otherwise. If site build errors with mkdocs-material is required, load add-python-package for mkdocs-material (agent) and build once more. Do not pixi add / uv add. If that skill is missing, or the retry still fails, name the error in one line. Name a build error; do not fail the fill, and do not claim the website was updated. When the build ran, say the website has been updated and paste the file:// line from that output verbatim. Do not name the markdown file or the HTML viewer path. Do not quote the CLI.

Stop conditions

Do not git commit or git end-turn.
Do not run cells run as a substitute for fill.
Do not pixi add / uv add; load add-python-package.
Missing skill or missing source → one-line skip.
Do not name --html or python -m skore_skills site build unless the user asked for HTML or a site viewer. Tell the user paths, not the CLI.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free