DOCSINSTALLATION & CLI

Getting Started

Install the SPM CLI, authenticate with your registry account, and build your very first autonomous agent skill.


1. Prerequisites

Before installing the Skill Package Manager CLI, ensure you have the following environments set up on your machine:

  • Node.js: Version 18.0 or higher.
  • Package Manager: npm, yarn, or pnpm.

2. Install the SPM CLI

The SPM CLI orchestrates your local skills registry development workflows, including downloads, auth, publishing, validation, and localized testing.

Install it globally using your preferred package manager:

bash
# Install globally via npm
npm install -g @agent-spm/cli

# Verify successful installation
spm --version

3. Authenticate with the Registry

To pull private packages or publish your own skills to the public SPM registry, you must link the CLI with your workspace registry tokens.

Run the following login wizard and follow the prompts to complete verification in your browser:

bash
# Launch the login prompt
spm login
Protip: If you are operating inside a headless CI/CD server, you can set the environment token directly using: export SPM_TOKEN="your_token_here".

4. Scaffold Your First Skill

Scaffold a completely compliant, boilerplate agent skill within a new subfolder using the spm init generator command:

bash
# Initialize a new skill package named "weather-checker"
spm init weather-checker

The CLI builds a beautifully structured schema bundle:

Generated Structure
weather-checker/
├── spm.yaml # Metadata & API configuration
├── index.ts # Main skill execution code
├── package.json # Node dependency requirements
└── README.md # Skill documentation and specs

5. Local Validation & Testing

You don't need a live LLM runtime pipeline to verify that your tool functions are formatted correctly. You can test your skill execution schemas locally via CLI mocking:

bash
# Validate spm.yaml metadata constraints
spm validate

# Run localized test execution against mock payloads
spm run --input '{"city": "San Francisco"}'