skill-authoring-standard

Defines quality standards for writing and reviewing SKILL.md files.

3|Updated Jun 15, 2026
One-click install
npx skills add https://github.com/patrickserrano/lacquer --skill skill-authoring-standard-patrickserrano
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-authoring-standard
Source: https://github.com/patrickserrano/lacquer/tree/main/core/skills/skill-authoring-standard
Command: npx skills add https://github.com/patrickserrano/lacquer --skill skill-authoring-standard-patrickserrano

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Skills often ship with vague descriptions, bloated bodies, or unnecessary companion files, making them hard to discover and useless when loaded. This standard gives authors and reviewers a concrete bar every SKILL.md must clear before it ships. ## Core Features & Use Cases - Frontmatter rules: Enforces exact name-to-directory matching, trigger-oriented descriptions, and disable-model-invocation for skills with real side effects. - Content discipline: Requires single responsibility per skill, instruction over exposition, and concrete checkable guidance instead of abstract best practices. - Placement and structure guidance: Decides when companion references/ or scripts earn their keep, and when guidance belongs in core/skills/, a profile's skills/, or path-scoped .claude/rules/ files. - Use Case: When writing a new skill for the lacquer or auditing the existing skill set, load this standard to check the description reads as a trigger, the body contains only actionable instructions, and no padding survives review. ## Quick Start Ask the AI to review your new SKILL.md against the skill-authoring-standard before shipping it.

Frequently Asked Questions about skill-authoring-standard

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a good SKILL.md description?▼

Write the description as a trigger, not a summary: state the scenarios, phrasings, or task shapes that should surface the skill. Keep the how-to in the body and limit the frontmatter to when a reader should load it.

How to review a Claude Code skill before shipping?▼

Check that the name matches the directory exactly, the description tells a first-time reader when to load it, every section gives an actionable instruction, and no companion file exists just for completeness. Cut rather than append when any check fails.

When should a skill use disable-model-invocation?▼

Use `disable-model-invocation: true` when a skill has real side effects such as shipping a release, pushing code, or deleting data. It ensures the skill only runs when a human explicitly types the slash command, never from model judgment.

When should I add references or scripts to a skill?▼

Add a `references/` directory only when material is long enough that inlining would bury the core instructions, and add a script only when the skill's value is executing it. A skill is a single SKILL.md by default.

Where should stack-specific guidance live instead of a skill?▼

Stack-specific rules that should load automatically when matching files are opened belong in path-scoped `.claude/rules/` files, not skills. A profile CLAUDE.md approaching 200 lines is the signal to split content into rules.