skill-creator

Automates creation, refinement, and evaluation of Claude skills.

29|89|Updated Dec 12, 2023
One-click install
npx skills add https://github.com/AxaFrance/design-system --skill skill-creator-axafrance
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/AxaFrance/design-system/tree/main/.github/skills/skill-creator
Command: npx skills add https://github.com/AxaFrance/design-system --skill skill-creator-axafrance

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Users often struggle to standardize the process of crafting Claude skills, track iterations, and capture reusable metadata. This Skill Unit provides a structured approach to drafting SKILL.md, organizing optional resources, and recording evaluation workflows so skills can be created and improved consistently.

Core Features & Use Cases

  • Structured skill creation: guides users to draft SKILL.md with required frontmatter (name + description) and a clear body that defines the workflow.
  • Iterative evaluation loop: supports running evals, collecting feedback, and refining skills across iterations.
  • Resource bundling: supports optional folders (scripts/, references/, assets/) that can be used by the skill when activated.
  • Metadata hygiene: ensures fields like dependencies and components are accurately captured for discovery and tooling.

Quick Start

Draft the SKILL.md with a clear name and description, then add optional resources and start the first evaluation cycle to begin iterating.

Frequently Asked Questions about skill-creator

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

FAQPage Schema
How do I automate Claude skill creation and ensure the frontmatter includes required metadata?

To automate Claude skill creation, draft a SKILL.md file with required frontmatter like name and description, then use the skill-creator unit to structure the workflow, bundle optional resources, and output a metadata package.

What's the best way to structure and refine an AI workflow with iterative evaluation?

The best way to refine an AI workflow is to run an iterative evaluation loop that executes benchmark tests, collects user feedback, and applies refinements across iterations to standardize skill performance.

Does skill-creation support bundling external scripts and reference assets into the metadata package?

Yes, skill-creation supports resource bundling by detecting optional folders such as scripts, references, and assets, and including them in the final structured metadata package for tooling and discovery.

How do I track dependencies and components metadata hygiene when designing new automation skills?

You track dependencies and components metadata hygiene by enforcing accurate capture of these fields during the skill creation process, ensuring the structured output passes toxicity checks and supports tooling discovery.

Can I benchmark performance and run evals on existing skills without starting from scratch?

Yes, you can benchmark performance and run evals on existing skills by applying the evaluation workflow to detect current resources, collect feedback, and iterate on the skill body without drafting a completely new SKILL.md.