# Example Skill

> A hand-written sample showing the shape yt2skill produces. Reach for the real thing when you need to turn a tutorial video or a manual into a skill an agent can follow.

- Skill: `mremanahmad/example-skill` (Agent Skill)
- Install (CLI): `npx skillmds@latest add mremanahmad/example-skill`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mremanahmad/example-skill/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: MrEmanAhmad (https://skillmd.com/u/mremanahmad)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/mremanahmad/example-skill

---


# example-skill

This file is not generated - it is a small, hand-written illustration of what a
yt2skill output looks like, so you know the shape before you run the tool.

A real generated `SKILL.md` has two parts:

1. **Frontmatter** (the `---` block above). `name` is the slug. `description` is the one
   line a session sees before it decides to load the skill, so it names the *trigger* and
   the *task*, not just the topic. "Reach for this when you need to X" beats "About X".

2. **The body**: what the thing is, then a procedure written as concrete actions the agent
   takes - real commands, real paths, real values. It never says "the video shows..."; it
   just gives the instruction, as if the author had learned the method and written it down.

## Example procedure

Say the source video demonstrated setting up a nightly database backup. The generated
skill would read something like:

1. Create the backup script at `/opt/backups/pg-backup.sh`:

   ```sh
   #!/usr/bin/env bash
   set -euo pipefail
   pg_dump -Fc mydb > "/opt/backups/mydb-$(date +%F).dump"
   find /opt/backups -name 'mydb-*.dump' -mtime +14 -delete
   ```

2. Make it executable: `chmod +x /opt/backups/pg-backup.sh`.

3. Add the cron entry (runs 02:30 nightly):

   ```
   30 2 * * * /opt/backups/pg-backup.sh
   ```

Note the exact retention window (14 days) and the exact time (02:30) - those numbers are
the product. A generated skill transcribes them precisely from the transcript or a frame,
and flags any it could not read rather than guessing.

Bulky detail (full config tables, option lists, edge cases) would go in a sibling
`reference.md`, keeping this file short enough to load and act on.

