# Your Skill Name

> Use this skill when <situation>. Covers <capabilities>. Use it whenever the user mentions <keywords>, asks to <phrasings>, or <adjacent situation> -- even if they don't explicitly say <the obvious term>.

- Skill: `ghostscientist/your-skill-name` (Agent Skill)
- Install (CLI): `npx skillmds@latest add ghostscientist/your-skill-name`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ghostscientist/your-skill-name/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- License: Apache-2.0
- Author: GhostScientist (https://skillmd.com/u/ghostscientist)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/ghostscientist/your-skill-name

---


# Your Skill Name

<!--
These six fields above are the COMPLETE set allowed by the Agent Skills spec
(https://agentskills.io/specification). Anything else -- `argument-hint`,
`model`, `context`, `paths`, `when_to_use`, etc. -- is Claude Code-specific and
will hard-error when packaging for claude.ai or the Skills API.

Run `python3 scripts/validate_skills.py` to check this file.
-->

One or two sentences on what this skill does and the failure mode it prevents.

## Workflow

1. Step one
2. Step two
3. Step three

## Step 1: ...

Write in the imperative, addressed to the agent. State what to do rather than
narrating why -- once a skill loads it stays in context, so every line is a
recurring token cost.

Only include what the agent would otherwise get wrong. If it already knows
something, cut it.

## Gotchas

- Concrete, non-obvious failure modes and their fixes.
- Avoid pinning versions, dates, or prices that will silently rot. Point at the
  authoritative URL instead.

<!--
Keep this body under 500 lines. Move detail into bundled files:

  your-skill-name/
  ├── SKILL.md       # this file
  ├── scripts/       # executable code the agent can run
  ├── references/    # docs loaded on demand
  └── assets/        # templates, images, data files

Reference them with relative paths, one level deep, and always say WHEN to
load each one:

  "Read `references/troubleshooting.md` if the build fails."   <- good
  "See references/ for more details."                          <- bad

Every bundled file must be referenced from SKILL.md (or from another reference
file). Unreferenced files can never be loaded by the agent; the validator
flags them as orphans.
-->