MLOps Prototyping CN

Structured Jupyter notebook prototyping with pipeline integrity

Security Scan

The skill does what it claims — a local notebook-structure checker — and its files and instructions are coherent and proportional to that purpose.

Purpose & Capability

Name/description match the included assets: SKILL.md documents a notebook-check workflow and the repo includes a small script (scripts/check-notebook.sh) that implements the check. No unrelated credentials, services, or installers are requested.

Instruction Scope

Runtime instructions only run a local shell script that opens the provided .ipynb and inspects cells. The script does not call external endpoints or read other filesystem paths, but it relies on python3 (not declared in metadata). The heuristics are simplistic (e.g., treating any digit in code as a 'magic number' unless RANDOM_STATE appears), and the outer shell prints a success message unconditionally after the Python block which may be misleading. No data exfiltration or broad system access is performed.

Install Mechanism

No install spec; instruction-only plus a small shipped script. Nothing is downloaded or extracted from remote URLs.

Credentials

No environment variables, credentials, or config paths are requested. The script only needs a local notebook file and a python3 runtime.

Persistence & Privilege

Skill does not request persistent/always-on presence, does not modify other skills or system settings, and contains no agents or background components.

Assessment

This skill appears coherent and low-risk: it runs a local script that reads a notebook file and prints checks. Before installing/running: (1) ensure python3 is available on the machine, (2) review scripts/check-notebook.sh yourself if you want to confirm behavior, (3) run it on non-sensitive copies of notebooks until you are comfortable with its heuristic false positives (its 'magic number' and import-detection heuristics are simplistic), and (4) be aware the wrapper prints a final 'Notebook check complete' message even when the internal check failed, so rely on the script's exit code rather than that message for CI gating.

MLOps Prototyping 🔬

Create standardized, reproducible Jupyter notebooks.

Features

1. Notebook Structure Check ✅

Validate notebook follows best practices:

./scripts/check-notebook.sh notebook.ipynb

Checks for:

• H1 title
• Imports section
• Config/Constants
• Data loading
• Pipeline usage

2. Template 📝

Use this structure:

1. Title & Purpose
2. Imports (standard → third-party → local)
3. Configs (all constants at top)
4. Datasets (load, validate, split)
5. Analysis (EDA)
6. Modeling (use sklearn.pipeline.Pipeline)
7. Evaluations (metrics on test data)

Quick Start

# Check your notebook
./scripts/check-notebook.sh my-notebook.ipynb

# Follow structure in notebook
# Use Pipeline for all transforms
# Set RANDOM_STATE everywhere

Key Rules

✅ DO:

• Put all params in Config section
• Use sklearn.pipeline.Pipeline
• Split data BEFORE any transforms
• Set random_state everywhere

❌ DON'T:

• Magic numbers in code
• Manual transforms (use Pipeline)
• Fit on full dataset (data leakage)

Author

Converted from MLOps Coding Course

Changelog

v1.0.0 (2026-02-18)

• Initial OpenClaw conversion
• Added notebook checker