Skip to content
Loop Engineering
Loop Engineering Tools

SKILL.md Generator for Loop Engineering

A skill is a reusable workflow package for an AI agent. A good SKILL.md tells the agent when to use the workflow, when not to, what steps to follow, what commands to run, and how to validate the result.

Create Reusable Loop Engineering Skills

Start from a preset or write your own. The SKILL.md is generated in your browser with YAML frontmatter and clear trigger, workflow, validation, and safety sections. This format is summarized from official provider guidance and kept vendor-neutral.

Need install paths? See SKILL.md Templates for Claude Code & Codex

Copy Your SKILL.md

Skill presets
SKILL.md
---
name: SEO Content Refresh
description: Refresh an existing SEO page while preserving search intent, URL, and internal links.
---
 
# SEO Content Refresh
 
## When to Use This Skill
Use when an existing ranking page needs updated facts, structure, or internal links without changing its target keyword or URL.
 
## When Not to Use This Skill
Do not use for brand-new pages, keyword research, or when the page's search intent is changing.
 
## Goal
Improve the page's usefulness and freshness while keeping its URL, title intent, and internal link strategy intact.
 
## Workflow
1. Read the current page and identify its primary search intent.
2. List outdated, thin, or missing sections.
3. Update content and FAQs without inventing facts.
4. Preserve or improve internal links.
5. Re-check title, H1, and meta description against intent.
 
## Validation
URL is unchanged.
Title and H1 still match search intent.
No invented statistics or citations.
Internal links preserved or improved.
 
## Safety Rules
Do not keyword stuff.
Do not invent data.
Do not change brand positioning.
Do not remove important internal links.
 
## Output Format
A revised page plus a short summary of what changed and why.
 
## References
Internal style guide and the existing page analytics.
 
## Stop and Escalate If
- The task requires permissions not available to the agent.
- The validation checks cannot be run.
- The workflow would require unsafe shortcuts.
- The agent is about to modify unrelated files or delete important artifacts.