Skip to content

getting-started.md

Latest commit

 

History

History
174 lines (119 loc) · 4.98 KB

getting-started.md

File metadata and controls

174 lines (119 loc) · 4.98 KB
Raw

Getting Started

This guide takes you from zero to your first skillet install in about five minutes.

Binary naming: the default command is skillet. sklt is provided as a shorthand alias — both names point to the same binary and accept the same arguments. All examples below use skillet; substitute sklt if you prefer the short form.

1. Install Skillet

Pick the install method that matches your platform. The npm package is getskillet (unscoped) with @echohello/skillet as a scoped alias; the binary is skillet (default) with sklt as an alias.

macOS / Linux (recommended)

curl -sSL https://echohello.dev/skillet/install.sh | sh

Or pick a specific channel:

Channel Command Notes
Homebrew brew install echohello-dev/tap/skillet macOS, Linuxbrew
npm (unscoped) npm i -g getskillet or use npx getskillet Node 20+
npm (scoped alias) npm i -g @echohello/skillet Same package, scoped name
Direct binary Download from GitHub Releases Single static binary, no runtime

Windows

irm https://echohello.dev/skillet/install.ps1 | iex
Channel Command Notes
winget winget install echohello-dev.skillet Windows 10+
Chocolatey choco install skillet
npm (unscoped) npm i -g getskillet Node 20+
npm (scoped alias) npm i -g @echohello/skillet Same package, scoped name

Container

docker run --rm ghcr.io/echohello-dev/skillet --help

Verify

skillet --version
# skillet/1.0.0 (commit=..., builtAt=..., target=...)

After global install, the sklt shorthand also works:

sklt --version
# skillet/1.0.0 (commit=..., builtAt=..., target=...)

2. Start a project

Create a new project directory and scaffold a manifest:

mkdir my-agent-skills && cd my-agent-skills
skillet init

This writes an apm.yml manifest declaring your project. Open it — it looks like this:

name: my-agent-skills
version: 0.1.0
description: Skills for my AI agent.
dependencies:
  apm: []

The format is OpenAPM v0.1 — Skillet is a conforming resolver. See Usage for the full schema.

3. Add your first skill

Point Skillet at any source that contains SKILL.md files:

# GitHub repo (uses git shorthand: owner/repo)
skillet add anthropics/skills

# Specific branch or tag
skillet add anthropics/skills#main

# OCI artifact from a registry
skillet add oci://ghcr.io/echohello-dev/frontend-design:v1

# Local directory
skillet add ../my-local-skills

skillet add resolves the source, discovers the skills inside it, and deploys them into your agent directory. By default it writes to whichever agent directory it finds (.claude/skills, .codex/skills, etc.). If none is found, it prompts you.

The source is appended to apm.yml so your install is reproducible.

4. Declare and install from a manifest

For reproducible installs across machines, commit an apm.yml and run:

skillet install

This reads every entry under dependencies.apm and installs it. No arguments required — apm.yml is the source of truth.

A realistic apm.yml for a team project:

name: my-agent-skills
version: 1.0.0
target: claude
dependencies:
  apm:
    - anthropics/skills/skills/frontend-design
    - git: https://gitlab.internal.acme.com/platform/coding-standards.git
      ref: v2.0
    - oci://ghcr.io/acme/security-checklist:v1.2.0
    - ./team-internal

See Usage for the full schema and source formats.

5. Verify

List the skills Skillet discovered in your project:

skillet find

Or search by keyword:

skillet find deploy

Inspect the lockfile (skillet.lock.yaml) to see what's installed, where it came from, and how it was placed:

cat skillet.lock.yaml

6. Update

When you want to refresh sources, edit the refs in apm.yml and re-run skillet install. For automated detection of newer refs:

skillet check    # show what would change
skillet update   # refresh refs and reinstall

(check and update are in-progress; today, bump refs in apm.yml manually and re-run skillet install.)

7. Iterate

Author a new skill in your project:

skillet init skills/team-voice

This scaffolds skills/team-voice/SKILL.md with the required frontmatter (name, description). Edit the body to define when the skill should trigger and what the agent should do.

Next steps

  • Read Usage for the full command reference, manifest schema, and source format details.
  • Read the APM Manifest ADR to understand which OpenAPM fields Skillet supports and which it intentionally ignores.
  • Read OCI Artifact Spec if you want to publish your own signed skill artifacts.