Getting Started¶

Welcome to AAM (Agent Artifact Manager) — the package manager for AI agent artifacts. Whether you're looking to install packages others have created, package your own skills and agents, or migrate existing artifacts, you're in the right place.

This section will get you from zero to productive in minutes.

Choose Your Path¶

Pick the path that best describes what you want to do:

Prerequisites¶

Before you begin, ensure you have the following installed:

What is AAM?¶

AAM is a package manager for AI agent artifacts — it's like npm for JavaScript or pip for Python, but for managing:

• Skills — Workflows and capabilities for AI agents (SKILL.md files with optional scripts, templates, references)
• Agents — AI agent definitions with system prompts, tool configurations, and skill references
• Prompts — Reusable prompt templates with variable interpolation
• Instructions — Platform-specific rules, coding standards, and guidelines

Why Use AAM?¶

Without AAM, managing AI agent artifacts means:

• ❌ Copy-pasting files between projects
• ❌ No versioning or dependency management
• ❌ Manual updates when something changes
• ❌ Platform-specific formats that don't transfer

With AAM, you get:

• ✅ Install packages with one command: aam install @author/python-best-practices
• ✅ Automatic dependency resolution: Skills that depend on other skills just work
• ✅ Multi-platform deployment: Write once, deploy to Cursor, Claude, Copilot, or Codex
• ✅ Version control: Use semver, pin versions, and share lock files with your team
• ✅ Local-first: Works offline with local registries — no server required

How AAM Works¶

flowchart LR
    Author["📦 Package Author"] -->|aam pkg publish| Registry["🗄️ Registry"]
    Registry -->|aam install| User["👤 User"]
    User -->|deploys to| Platforms["🤖 AI Platforms<br/>(Cursor, Claude, etc.)"]

    style Author fill:#e3f2fd
    style Registry fill:#f3e5f5
    style User fill:#e8f5e9
    style Platforms fill:#fff3e0

1. Authors create packages — Bundle skills, agents, prompts, and instructions into a package with aam.yaml
2. Packages are published — Upload to a registry (local filesystem or HTTP server)
3. Users install packages — aam install downloads, resolves dependencies, and deploys artifacts
4. Artifacts are deployed — AAM places files in the right location for your AI platform

Section Contents¶

This Getting Started section contains:

Next Steps¶

Ready to get started? Pick your path:

• New to AAM? Start with Installation to get AAM set up
• Want to see it work fast? Jump to Quick Start for a 5-minute demo
• Ready to build? Go to Your First Package to create a complete package