skill-authoring

Validates and standardizes SKILL.md files against repository authoring conventions.

Updated Apr 12, 2026
One-click install
npx skills add https://github.com/azborgonovo/ai-skills --skill skill-authoring-azborgonovo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-authoring
Source: https://github.com/azborgonovo/ai-skills/tree/main/.claude/skills/skill-authoring
Command: npx skills add https://github.com/azborgonovo/ai-skills --skill skill-authoring-azborgonovo

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Writing SKILL.md files that Claude parses correctly requires following many subtle formatting and retrieval conventions, and mechanical rules like the description length cap or the two---- rule drift silently until a full review pass catches them. ## Core Features & Use Cases - Authoring conventions: Codifies rules for horizontal rules, bold usage, line wrapping, American spelling, description phrasing ("Use when ..."), cross-skill references, and negative instructions. - Eval-driven verdicts: Defines thresholds for keeping, improving, promoting, or discarding a skill based on measured pass-rate deltas, pass@k versus pass^k, and the Fired column. - Bundled validator: Ships scripts/check_skills.py, which checks name charset, description length, frontmatter delimiters, body length, and hard-wrapped prose across any skill tree. - Use Case: Before committing a new skill, run the validator over the repo and apply the conventions to fix a description that exceeds the 1024-character cap. ## Quick Start Review my new SKILL.md against this repo's authoring conventions and run the bundled check_skills.py validator on it.

Frequently Asked Questions about skill-authoring

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

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

Write the description as retrieval text in third person: state what the skill does, then state when to use it with "Use when the user ..." phrasing. Keep it under 1024 characters and quote phrasings a real request carries.

How do I validate SKILL.md files before committing?▼

Run python .claude/skills/skill-authoring/scripts/check_skills.py with no arguments to check skills/ and .claude/skills/, or pass a path to check a tree elsewhere. It exits non-zero on failures and names the file and rule.

What rules does check_skills.py enforce?▼

It checks only machine-settleable rules: name charset and length, description length cap, exactly two frontmatter delimiters, body line limit, and hard-wrapped prose. Judgment calls like no-op sentences or third-person voice stay with the reviewer.

When should a skill be discarded based on eval results?▼

Discard a reasoning skill when the no-plugin baseline matches or beats the with-plugin score on cases where the skill fired. Read three runs, check the Fired column first, and use pass^k for skills that must work every time.

Can a SKILL.md reference skills from other plugins?▼

Reference sibling skills in the same plugin freely since they install together. Avoid referencing skills in other plugins because they install independently and may be absent; phrase any such help as an optional accelerator with a local fallback.