optimizely-content-modelling

QA Skills

A collection of AI skills for QA workflows — test generation, refinement docs, design verification, and more.

Skills come in two flavours:

Type	Description	Released as .skill?
Distributable	Web-based skills for Claude.ai	Yes
IDE-only	Skills for GitHub Copilot or local development use	No

---

Skill structure

Each skill lives in skills/<skill-name>/ and must contain a SKILL.md with YAML frontmatter:

---
name: my-skill
description: When and how to use this skill.
---

To mark a skill as distributable, add an empty .distribute file to its folder:

skills/my-skill/
├── SKILL.md
├── .distribute       ← presence of this file triggers packaging
└── ...

Skills without a .distribute file are ignored during packaging. The .distribute file is never included in the packaged .skill archive.

---

Releases

Releases are created by pushing a version tag. The GitHub Actions workflow checks out the committed code, packages every distributable skill into a .skill file (a zip with a .skill extension), and attaches them to a GitHub Release.

git tag v1.0.0
git push origin v1.0.0

The workflow then appears under Actions → Release Skills and the .skill files are available as downloadable assets on the release.

---

Adding a skill to claude.ai (Claude for Web)

1. Download the relevant skills from the latest release.
2. Go to skills in Claude.ai
3. Upload the .skill file

---

Local packaging

To package a skill locally (for testing):

bash scripts/package-skill.sh skills/<skill-name>

# Output to a specific directory
bash scripts/package-skill.sh skills/<skill-name> dist/

This does not affect releases — releases always build from committed code in CI.

---

Adding a new skill

1. Create a folder under skills/<skill-name>/
2. Add a SKILL.md with the required frontmatter
3. Add an empty .distribute file to the folder if it should be included in releases
4. Commit and push — tag a new version when ready to release