# Example Skill

> Template skill demonstrating the shared SKILL.md format. Replace with a real skill; use when asked to demonstrate how skills in this repo work.

- Skill: `iancchambers/example-skill` (Agent Skill)
- Install (CLI): `npx skillmds@latest add iancchambers/example-skill`
- Raw SKILL.md: https://api.skillmd.com/api/skills/iancchambers/example-skill/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: iancchambers (https://skillmd.com/u/iancchambers)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/iancchambers/example-skill

---


# Example Skill

This is a placeholder showing the structure every skill in this repo follows.

## Format

- Each skill lives in `skills/<skill-name>/` with a `SKILL.md` at its root.
- Frontmatter requires `name` (kebab-case, matching the folder) and
  `description` (one line telling the agent *when* to use the skill — this is
  what the agent reads to decide relevance, so make it specific).
- The body contains the actual instructions the agent follows once the skill
  is invoked.
- Supporting files (scripts, references, templates) can sit alongside
  `SKILL.md` in the same folder and be referenced by relative path.

## Optional control fields

Beyond `name`/`description`, frontmatter can carry control fields (shown
commented out above):

- `allowed-tools` — allowlist of tools the agent may use while the skill runs
  (e.g. `Read, Grep, Bash(git *)`). Use for skills that should never write or
  run arbitrary commands.
- `disable-model-invocation: true` — the skill only runs when the user invokes
  it explicitly (`/skill-name`); the model won't trigger it on its own. Use
  for consequential actions like deploys or releases.
- `license` — license for the skill content.
- `metadata` — free-form key/value map (version, author, tags) for your own
  tooling; agents don't act on it.

Portability: unknown frontmatter fields are ignored, so these are safe to
include everywhere — but as of now only Claude Code reliably *enforces*
`allowed-tools` and `disable-model-invocation`. Don't rely on them as a
security boundary in other agents; keep the skill body's instructions safe on
their own.

## Instructions

When invoked, explain to the user how skills in this repo are structured and
point them at the README for how to add a new one.