Skip to content

Invocation Control & Arguments

By default both you and the model can invoke a skill. Two frontmatter switches change that:

Frontmatter Who can invoke Use for
(default) You and the model Most skills: conventions, procedures, reference knowledge
disable-model-invocation: true Only you, via /name Side effects: /deploy, /commit, /release; the model never decides to run these on its own
user-invocable: false Only the model Background knowledge, hidden from the / menu

The editor shows a user-only /deploy skill. Note the combination: a skill the model can’t trigger, with steps too fragile to leave to judgment.

  • $ARGUMENTS: everything typed after the command
  • $0, $1: positional; named arguments via an arguments: list, with argument-hint shown in the menu
  • ${CLAUDE_SKILL_DIR}: absolute path to the skill’s folder, and how you reference bundled scripts

!`command` runs before the model reads the prompt and replaces the placeholder with its output; you used it in /standup. Multi-line blocks use a fenced ```! block.

Prescriptive, exact commands for fragile sequences (“Run exactly this; do not modify”). Freedom plus reasons for flexible tasks. Provide a default, not a menu: name one library, mention the fallback, and don’t survey five.