Skill creator

Skill aizech/bernhard-zechmann-skills/skills/engineering/skill-creator

Modular AI agent skills used across engineering, writing, and research workflows. Model‑agnostic, composable, and production‑tested. Includes engineering, productivity, personal, and experimental skill packs.

Install

npx -y skills add aizech/bernhard-zechmann-skills --skill skill-creator

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Create a new skill or improve an existing one. Use when the user wants to capture a workflow as a reusable skill, edit a skill, or optimize its triggering.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.2 KB, as published. Nobody here has run it

Skill Creator

Capture a workflow as a reusable skill, then test and tighten it.

Process

1. Capture intent

Ask:

What should the skill enable the agent to do?
When should it trigger? Which phrases or contexts?
What is the expected output format?
Should we add test cases? (Yes for deterministic outputs; optional for subjective ones.)

2. Write the SKILL.md

Create a folder with the skill name and a SKILL.md:

---
name: skill-name
description: When to trigger and what it does. Be specific.
license: MIT
compatibility: claude-code opencode github-copilot devin pi cursor
---

Body rules:

One responsibility per skill.
Imperative instructions.
Include examples only when they clarify.
Keep it under 250 lines. Split into references/ if it grows.
No emojis or Material icons.

3. Create test cases

Write 2–3 realistic prompts a user would actually say. Save them to evals/evals.json:

{
  "skill_name": "example-skill",
  "evals": [
    {
      "id": 1,
      "prompt": "User's realistic task",
      "expected_output": "Description of expected result",
      "files": []
    }
  ]
}

4. Run and evaluate

For each test case:

Run the skill against the prompt.
Run a baseline without the skill if needed.
Capture outputs and timing.
Review with the user. Use a viewer or inline comparison when subagents are not available.

5. Iterate

Fix the biggest failure first.
Remove instructions that do not pull their weight.
Explain the why, not just the what.
Avoid overfit changes that only help the test cases.

6. Optimize the description

When the skill is stable, tune the description frontmatter so the agent triggers it reliably on real prompts. Use a mix of should-trigger and should-not-trigger examples.

Reference

Progressive disclosure: metadata, SKILL.md body, bundled resources.
Keep bundled resources in scripts/, references/, or assets/.