skill-writer

Design contract-first Claude skills with JSONL-first prompts and progressive disclosure.

1|Updated Feb 17, 2026
One-click install
npx skills add https://github.com/lprior-repo/claude-skills --skill skill-writer-lprior-repo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-writer
Source: https://github.com/lprior-repo/claude-skills/tree/main/skill-writer
Command: npx skills add https://github.com/lprior-repo/claude-skills --skill skill-writer-lprior-repo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Meta skill for authoring Claude skills using contract-first design, JSONL-first prompts, and progressive disclosure docs. Use when creating or improving SKILL.md files.

Core Features & Use Cases

  • Establishes a contract-driven structure for skill creation, ensuring consistent frontmatter and lean execution blocks.
  • Encourages progressive disclosure by moving long guidance to support docs and linking them from SKILL.md.
  • Provides templates, references, and examples to accelerate design, review, and refactoring of SKILL.md files.

Quick Start

Draft a new skill named <skill-name> using contract-first design with JSONL-first prompts and progressive disclosure; place heavy guidance in support docs and link them from SKILL.md.

Frequently Asked Questions about skill-writer

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

FAQPage Schema
What is contract-first Claude skill design?

Contract-first Claude skill design establishes a contract-driven structure for creating SKILL.md files, ensuring consistent frontmatter fields and lean JSONL execution blocks before adding detailed guidance.

How do I write a SKILL.md file using progressive disclosure?

To write a SKILL.md file using progressive disclosure, keep the JSONL behavior block lean by moving heavy guidance into linked support docs, ensuring rapid iteration and safe defaults.

Can I use this approach to refactor existing AI skills?

Yes, you can apply contract-first design to refactor existing AI skills by enforcing explicit frontmatter fields and restructuring prompts into a lean JSONL behavior block.

What's the best way to structure JSONL-first prompts for Claude skills?

The best way to structure JSONL-first prompts is to define a lean JSONL behavior block within the SKILL.md file, enforcing explicit frontmatter and linking to external support docs for heavy guidance.

Why does my SKILL.md file need explicit frontmatter fields?

Your SKILL.md file needs explicit frontmatter fields to enforce consistent contract-driven structure and safe defaults, ensuring the AI skill behaves predictably during rapid iteration.

When should I move guidance out of my SKILL.md file?

You should move heavy guidance out of your SKILL.md file into linked support docs when applying progressive disclosure, keeping the core JSONL behavior block lean for faster processing.