Short answer
An AI package manager helps teams install, update, and share reusable AI setup the way npm does for code. AIPM installs versioned agent skills (SKILL.md folders) for Claude Code, Codex, and Cursor, and tracks AI prompts as Markdown snapshots. Rules, MCP servers, and hooks are planned, not installable today.
Words used in this guide
- AIPM: a tool that installs and updates reusable AI instructions.
- Package: a named group of files that people can install and update together.
- Workflow: a set of steps used to finish a task.
- Git: a tool that records file changes so people can review or undo them.
- Config: settings that tell a tool how to work.
- MCP: a standard way for an AI tool to connect to other tools and information.
- CLI: a tool that you use by typing commands in a terminal.
What this means
The simple idea
Normal software teams use package managers to install code. AI teams also need reusable files, but the files are different: skill folders with a SKILL.md, examples, scripts, and prompts. An AI package manager gives those files a name, a version, and a clear install flow.
Why it matters
Without a package manager, people copy prompts from chat, paste rules by hand, and forget which version is current. That works for one person, but it breaks when a team grows.
How AIPM fits
AIPM gives you a registry and a CLI. You find a skill, run aipm add with an exact version, and AIPM writes the skill folder into the project: .claude/skills/<skill>/ for Claude Code (--target claude) or .agents/skills/<skill>/ for Codex (--target codex). Cursor also loads skills from both folders, so either target works in Cursor. The version is recorded in aipm.package.json and aipm-lock.json.
What AIPM installs today
Skills are the only installable package type. Prompts from the AIPM prompt library can be tracked with aipm add <prompt URL>, which saves a Markdown snapshot in .aipm/prompts/; prompts do not have version numbers. Rules, MCP server config, hooks, and other tool config files are planned but are not installable package types yet, so keep them in Git for now.
Team workflow
Commit aipm.package.json and aipm-lock.json. Teammates run aipm install to get the same pinned versions, aipm update moves a skill to its latest version, and aipm remove cleans it up. Private org packages work after aipm login, and CI can use an org install token. Pin exact versions: version ranges are not supported.
Simple steps
- Find a skill in the AIPM registry.
- Install the AIPM CLI.
- Run aipm init in your project.
- Run aipm add @scope/name@version --target claude (or --target codex).
- Open Claude Code, Codex, or Cursor and use the installed skill.
Where to go next
If you want to try this in a real project, start with how to use AIPM and the command reference. If you want to share your own workflow, read the publishing guide.
FAQ
Is an AI package manager the same as npm?
No. npm installs code packages. AIPM installs versioned AI agent skills and tracks AI prompts. It does not install rules, MCP config, or hooks today.
Who should use an AI package manager?
Developers and teams who use AI tools in more than one project should use one. It helps keep setup repeatable.
Which AI tools does AIPM install skills into?
Claude Code (--target claude writes .claude/skills/<skill>/SKILL.md) and Codex (--target codex writes .agents/skills/<skill>/SKILL.md). Cursor loads skills from .claude/skills and .agents/skills too, so use either target for Cursor.