Skill template with strong triggers

A parameterized SKILL.md skeleton following the skill-creator methodology: third-person description packed with concrete trigger phrases, bash injection for live context, $ARGUMENTS parameterization, and progressive-disclosure pointers to bundled references.

Parameters

Parameter	What to supply
`<skill-name>`	kebab-case folder name; becomes /<skill-name>
`$ARGUMENTS`	everything typed after the skill name
`$0, $1…`	positional arguments
`${CLAUDE_SKILL_DIR}`	path placeholder for bundled scripts

The artifact

---
description: This skill should be used when the user asks to "<trigger phrase 1>", "<trigger phrase 2>", "<trigger phrase 3>", or mentions <domain keywords>. Provides <one-line summary of what the skill delivers>.
argument-hint: <what-to-type-after-the-name>
# user-only workflow with side effects? uncomment:
# disable-model-invocation: true
# background knowledge Claude applies silently? uncomment:
# user-invocable: false
---

# <Skill Title>

## Context to gather first

!`<shell command that injects live context, e.g. git diff $ARGUMENTS>`

## Instructions

Do <the task> for: $ARGUMENTS

1. <Step one — imperative form, specific not generic>
2. <Step two — anticipate the common edge case>
3. <Step three — define the output format explicitly>

## Quality bar

- <A checkable acceptance criterion>
- <Another criterion>

## Going deeper

For detailed patterns consult the bundled references:
- references/<topic>.md — <what it covers>
- examples/<example>.md — <working example>
- Run scripts with: bash ${CLAUDE_SKILL_DIR}/scripts/<tool>.sh

Keep this file lean (~1,500–2,000 words). Push detail into references/
so it loads only when needed — progressive disclosure.