CLI
arctl skill publish

arctl skill publish

Publish a skill to the registry.

This command supports multiple modes:

  1. From a local skill folder (with SKILL.md): Reads metadata from the local file and publishes with a Git repository reference or a pre-built Docker image reference.
  2. Direct registration with Git (without SKILL.md): Registers a skill by name using --git and --version flags.
  3. Direct registration with Docker (without SKILL.md): Registers a skill by name using --docker-image and --version flags.

Usage

arctl skill publish <skill-name|skill-folder-path> [flags]

Examples:

# Publish from a local folder with a Git repository reference
arctl skill publish ./my-skill --git https://github.com/myorg/repo --version 1.0.0

# Publish from a local folder with a pre-built Docker image
arctl skill publish ./my-skill --docker-image docker.io/myorg/my-skill:v1.0.0 --version 1.0.0

# Register directly from a Git repository (no local files needed)
arctl skill publish my-skill \
  --git https://github.com/myorg/repo/tree/main/skills/my-skill \
  --version 1.0.0 \
  --description "My remote skill"

# Register directly with a pre-built Docker image (no local files needed)
arctl skill publish my-skill \
  --docker-image docker.io/myorg/my-skill:v1.0.0 \
  --version 1.0.0 \
  --description "My Docker skill"

# Preview what will be published without publishing the skill
arctl skill publish ./my-skill --git https://github.com/myorg/repo --version 1.0.0 --dry-run

Command-specific flags

--description string    Skill description (optional, used with direct registration)
--docker-image string   Docker image URL (e.g., docker.io/myorg/my-skill:v1.0.0)
--dry-run               Show what would be done without actually doing it
--git string            Git repository URL. Supports tree URLs: https://github.com/owner/repo/tree/branch/path
--version string        Version to publish (required for --git or --docker-image)
The --git and --docker-image flags are mutually exclusive. One of them is required.
To build a skill as a Docker image, use arctl skill build instead.

Global flags

-v, --verbose: Enable verbose output.
-h, --help: Display help information for the command.