agentleFS
jsboigejsboige/CoursIA/.claude/skills/enrich-notebooks/SKILL.md

enrich-notebooks

Skill17 starsChanged 2 days agoMITView on GitHub

Enrich Jupyter notebooks with pedagogical markdown content. Arguments: [target] [--execute] [--fix-errors] [--strict] [--consecutive] [--iterate]

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Add pedagogical markdown content to Jupyter notebooks.

    line 8
  2. Very specific:

    Read the notebook-enricher agent instructions (.claude/agents/notebook-enricher.md)

    line 25
  3. Very specific:

    Verify with: helper.get_cell_sequence(start, end) after each insertion

    line 36
  4. Very specific:

    If --iterate: Use notebook-cell-iterator agent for targeted cell fixes

    line 41
  5. Very specific:

    Use model: sonnet for enrichment agents (good balance of speed and quality).

    line 45
  6. Distinctive:

    Verify with git diff (expect more insertions than deletions)

    line 70
  7. Very specific:

    Consider running /cleanup-notebooks to fix any positioning issues

    line 79
  8. Distinctive:

    Update MEMORY.md with lessons learned about positioning accuracy

    line 80

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Enrich Notebooks

Add pedagogical markdown content to Jupyter notebooks.

Target: $ARGUMENTS

Arguments

  • target: Notebook path, family name (Infer, Sudoku, Tweety, Lean, GenAI, etc.), or all
  • --execute: Run notebooks and capture outputs before enriching
  • --fix-errors: Correct code errors found during execution
  • --strict: Require interpretation after EVERY code cell
  • --consecutive: Focus on consecutive code cells without markdown
  • --iterate: Use cell-iterator for iterative correction

Process

  1. Parse target - Discover notebooks to enrich
  2. For each notebook, launch a background agent:
    • Read the notebook-enricher agent instructions (.claude/agents/notebook-enricher.md)
    • Analyze structure: python scripts/notebook_tools/notebook_helpers.py list {path} --verbose
    • Find gaps programmatically (Python API):
      from scripts.notebook_helpers import NotebookHelper
      helper = NotebookHelper(path)
      gaps = helper.find_cells_needing_enrichment()      # Cells missing interpretation
      consecutive = helper.find_consecutive_code_cells()  # Code cells without markdown between
    • Validate placement: helper.validate_enrichment_context(cell_index) returns what's needed
    • Insert markdown cells via NotebookEdit (bottom-to-top to preserve indices)
    • Verify with: helper.get_cell_sequence(start, end) after each insertion
  3. If --execute: Execute notebooks first to capture outputs for interpretation
  4. If --fix-errors: Analyze errors, propose corrections, re-execute
  5. If --consecutive: Prioritize fixing consecutive code cells
  6. If --iterate: Use notebook-cell-iterator agent for targeted cell fixes

Agent Delegation

Use model: sonnet for enrichment agents (good balance of speed and quality). For complex domains (Probas, Lean), consider model: inherit for better reasoning.

Task(
    subagent_type="general-purpose",
    model="sonnet",
    prompt="Tu es un agent notebook-enricher. Lis .claude/agents/notebook-enricher.md. Enrichis: {path}",
    description=f"Enrich {name}",
    run_in_background=True
)

Enrichment Criteria

TypePlacementTense
Section introBEFORE codeFuture: "This code will..."
Code explanationBETWEEN code cellsPresent: "This function..."
Result interpretationAFTER code outputPast: "The results show..."
TransitionBetween sections"After seeing X, let's explore Y..."
ConclusionEnd of sectionSummary table

After Enrichment

  • Verify with git diff (expect more insertions than deletions)
  • Run the enrich-quality scanner (rule C.7, mandatory):
    git show origin/main:<nb-path> > /tmp/base.ipynb
    python scripts/notebook_tools/scan_enrich_quality.py <nb-path> --base /tmp/base.ipynb
    # 0 HIGH finding before commit: anchors must resolve at HEAD (code[N] =
    # N-th CODE cell, 0-based, final layout), accents must survive, hrefs must
    # resolve, no phantom entities, no worked solution before a TODO exercise.
  • Consider running /cleanup-notebooks to fix any positioning issues
  • Update MEMORY.md with lessons learned about positioning accuracy

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.