Deploy Artifacts (CLI)¶

See how to deploy artifacts from your collection into a project's .claude/ directory, making them instantly available to Claude Code.

About This Demo

Duration: ~45 seconds
Audience: Developers bootstrapping projects
What you'll see: Collection inventory, empty project state, deploying an artifact, verifying deployment

Deploy Artifacts CLI Demo

What You'll See¶

Collection Inventory¶

Start by reviewing what's available in your collection.

skillmeat list --collection my-collection

What's happening: - Your collection holds multiple deployable artifacts: skills, commands, agents, and more - All are ready to deploy into any project - The collection is your personal library — nothing is project-specific yet

Empty Project State¶

Check the target project before deployment.

ls -la ./my-project/.claude/

What's happening: - The .claude/ directory exists with empty subdirectories - skillmeat.toml is the project manifest — it will be updated by deploy - This is a clean slate — no artifacts deployed yet - A good starting point to see the change after deployment

Deploying an Artifact¶

Deploy a single artifact from your collection to the project with one command.

skillmeat deploy <artifact-name> \
  --collection my-collection \
  --project ./my-project \
  --type skill \
  --overwrite

What's happening: - --collection selects the source artifact - --project selects where to deploy - --type skill resolves ambiguity when the same name exists in multiple types - --overwrite skips the interactive confirmation — essential for scripts - No prompts or confirmation dialogs — immediate deployment

Collection Unchanged After Deploy¶

Your collection remains untouched — deploy is non-destructive.

skillmeat list --collection my-collection

What's happening: - The same four artifacts appear — nothing was removed or modified - Deploy reads from the collection; it doesn't consume - The same artifact can be deployed to many projects from one collection - Re-running deploy with --overwrite is safe and idempotent

Verifying Deployment¶

Check that the artifact landed in the project's .claude/ directory.

ls ./my-project/.claude/skills/

What's happening: - Your deployed artifact appears in the subdirectory matching its type - Claude Code finds it automatically in .claude/ — no extra configuration - This is the same path you'd create manually; SkillMeat just manages it - Ready to use immediately

Deploy vs. Scaffold¶

Both commands populate a project, but they work differently:

Command	What It Does	When to Use
scaffold	Analyzes PRD/README, auto-detects stack, populates project	Starting a greenfield project
deploy	Copy specific artifact from collection to project	Adding individual artifacts to existing project

Use scaffold to bootstrap; use deploy to add more.

Key Takeaways¶

One-command deployment: Point to source and destination, deploy is instant
Non-destructive: Collection is unchanged after deployment
Reusable: One collection can deploy to many projects
Idempotent: Safe to re-run with --overwrite
Type resolution: --type flag prevents ambiguity

Try It Yourself¶

# List what you have in your collection
skillmeat list

# Initialize a new project
mkdir my-project
cd my-project
skillmeat init --project . --type project

# Deploy a skill from your collection
skillmeat deploy <skill-name> --collection my-collection --project . --type skill --overwrite

# Verify it's there
ls -la ./.claude/skills/

# Deploy more artifacts as needed
skillmeat deploy <command-name> --collection my-collection --project . --type command --overwrite

Common Flags¶

Flag	Purpose
`--collection <name>`	Source collection (default: active)
`--project <path>`	Target project directory
`--type <type>`	Artifact type: skill, command, agent, etc.
`--overwrite`	Overwrite without confirmation (safe for scripts)
`--force`	Force deployment (use with caution)