Hello World

Hello World

The example skill of this repository: a meta skill that explains the repo's own contract. It exists as a smoke test for the install/validate/CI pipeline — if this skill lints and is discovered, the contract machinery works.

When to use

Trigger this skill when the user asks how this repository works: what a SKILL.md must contain, how skills are discovered and installed, what skills-ref checks, or how to add a new skill. Do not use it for questions about any other skill's domain.

Quick start

1. Read the contract: AGENTS.md — it is the binding rulebook.
2. To add a new skill: ./scripts/new-skill.sh <kebab-case-name> (use --dry-run first to preview).
3. Verify any skill locally: skills-ref validate skills/<name>/.
4. Install skills from this repo: npx skills add <owner>/<repo>.

Workflow

1. A contributor scaffolds a skill with scripts/new-skill.sh <name>.
2. They replace every <!-- TODO: ... --> marker in the generated SKILL.md.
3. They run skills-ref validate skills/<name>/ — it enforces the frontmatter contract (name matches directory, valid description, no forbidden fields).
4. CI (lint-skills.yml) runs the validator per skill and npx skills@latest add . --list to prove the repo is discoverable.
5. Users install via npx skills add <owner>/<repo>.

If validation fails, the error names the field and the rule — fix it, re-run, done.

Examples

# Scaffold a new skill (prints actions, writes nothing)
./scripts/new-skill.sh pdf-extract --dry-run

# Validate an existing skill
skills-ref validate skills/hello-world/

# Install from the published repo
npx skills add <owner>/<repo> --skill hello-world

References

• references/glossary.md — definitions of contract terms (frontmatter, progressive disclosure, .template/, skills-ref); read when a term in the contract is unclear.