subskill-generation-rule_skill

Overview

This skill defines a simple, enforceable folder layout and placement rules for generated subskills and artifacts. It standardizes where recommendation outputs, generated scripts, and feature-level instruction files belong to keep the main skill root uncluttered. The goal is predictable organization that scales with multiple features and versions.

How this skill works

The skill inspects how generated files are placed and indicates the correct destinations: data/ for recommendation or result artifacts, and subskills/<feature>/ for new feature scripts. It enforces one folder per feature and expects an optional feature-level instruction file inside each feature folder when usage or behavior needs documentation. It flags or prevents one-off scripts and generated files from being placed in the main skill root.

When to use it

• Creating a new subskill or feature implementation
• Adding generated recommendation outputs or result artifacts
• Refactoring or organizing existing features into canonical folders
• Enforcing consistent file placement across team contributions
• Preparing a project for archiving or backup

Best practices

• Always write generated recommendation outputs to data/ to separate artifacts from source code.
• Create one folder per feature under subskills/ to keep features isolated and discoverable.
• Place new feature scripts inside subskills/<feature>/, not in the project root.
• Include a short feature-level instruction file inside the feature folder when usage, inputs, or behavior are nontrivial.
• Avoid committing one-off scripts or generated files to the main skill root to reduce clutter and merge conflicts.

Example use cases

• Add a new recommendation engine: place scripts in subskills/recommender/ and outputs in data/
• Refactor a mixed root directory: move scripts into subskills/<feature>/ and results into data/
• Onboarding contributors: provide feature-level instruction files inside each subskills folder to explain setup and behavior
• Automated CI: validate that no generated artifacts or feature scripts appear in the root before merging
• Archiving releases: collect versioned artifacts from data/ and feature code from subskills/ for backup