skills.rest
thensls
thenslsOFFICIAL
Explore all skills by this author →
WebsiteGitHub

skill-creation

Codify a design rubric for reusable conversational skills with safety tiers and diagnostic loops.

6|Updated Mar 26, 2026
Search Tags:#evaluation#safety#rubric#cross-system#skill-creation#skill-design#diagnostic-loop
One-click install
npx skills add https://github.com/thensls/nsls-builder-toolkit --skill skill-creation-thensls
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-creation
Source: https://github.com/thensls/nsls-builder-toolkit/tree/main/skills/skill-creation
Command: npx skills add https://github.com/thensls/nsls-builder-toolkit --skill skill-creation-thensls

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Prevents teams from re-discovering the same design and safety patterns when building conversational skills by capturing the fractal rubric — safety tiers, purpose, diagnostic loops, domain micro, and cross-system awareness — while the experience is still fresh.

Core Features & Use Cases

  • Three-tier safety model: Defines read-only, configuration, and destructive tiers so permission boundaries are explicit from the start.
  • Trigger-first frontmatter: Shows how to write a description that determines when the skill should be invoked rather than how to use it.
  • Diagnostic loop & domain micro: Encourages TRY → OBSERVE → DIAGNOSE → ADAPT cycles and capturing hard-won gotchas from real usage.
  • Creation flow: Guides builders through design, eval, and quality-gate phases so skills ship with measurable reliability.
  • Use case: A developer who just integrated PostHog or Slack can capture exact config gotchas, draft a SKILL.md, and run evals to ensure reliable triggering.

Quick Start

Draft a SKILL.md with name and a trigger-focused description, add a top-level safety block, write a one-paragraph purpose, and include domain micro and a diagnostic loop before saving.

Frequently Asked Questions about skill-creation

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

FAQPage Schema
How do I create reusable agent skills with safety tiers and diagnostic loops?▼

To create reusable agent skills, draft a SKILL.md with explicit frontmatter, a top-level safety model defining read-only and destructive tiers, and a TRY → OBSERVE → DIAGNOSE → ADAPT loop to capture real usage gotchas.

What is a trigger-first description when designing conversational skills?▼

A trigger-first description is frontmatter text that determines when a conversational skill should be invoked rather than how to use it, ensuring reliable automated triggering based on user intent.

How do I structure a SKILL.md file for Claude Code automations?▼

Structure a SKILL.md by starting with explicit name and description frontmatter, adding a first-section safety block, writing a one-paragraph purpose statement, and including domain micro details and a diagnostic loop.

Can I use this skill creation rubric for cross-system awareness and integrations?▼

Yes, the skill creation rubric explicitly supports cross-system awareness by cross-referencing setup with /connect and /data-intel, allowing you to capture configuration gotchas from integrations like PostHog or Slack.

What's the best way to evaluate conversational skills before shipping?▼

The best way to evaluate conversational skills is to run them through design, eval, and quality-gate phases, ensuring the skill ships with measurable reliability and proper trigger guidance.

Why do I need a three-tier safety model for agent-driven automations?▼

A three-tier safety model defines read-only, configuration, and destructive tiers to make permission boundaries explicit from the start, preventing unauthorized destructive actions during agent-driven automations.

Related Skills

Explore More Software Engineering Skills→
Norman-buryNorman-bury
3.0k

environment-setup

Install Miniconda and configure a reproducible Python data-visualization environment.

Community
Advanced
coreyjacoreyja
12

skill-writer

Draft SKILL.md frontmatter and structure guidance for Claude Code skills.

Community
Basic
PostHogPostHog
58

integration-react-vite

Integrate PostHog analytics into React apps built with Vite.

Official
Intermediate
slowman2084slowman2084
6

meta-ideal-state

Generate structured ideal_state.md documents defining Agent goals, intents, safety rules, and output templates.

Community
Advanced