What is mpak?

mpak is a package registry for distributing capabilities that make AI agents smarter. Think npm for AI agents.

mpak distributes MCP (Model Context Protocol) servers and Agent Skills. MCP is an open standard that lets AI agents connect to external tools and data sources through a unified interface.

# Search for MCP bundles
mpak bundle search postgres

# Run an MCP server instantly
mpak bundle run @nimblebraininc/echo

No manual dependency installation. No configuration. Just works.

The Problem

AI agents are generalists. They become powerful when you give them specialized tools and knowledge. But distributing these capabilities is a mess:

MCP servers require manual dependency management and complex setup
Skills are copy-pasted between projects with no versioning
Discovery means searching GitHub and hoping you find something that works

Registries solved this problem for every other ecosystem: npm for JavaScript, PyPI for Python, crates.io for Rust. mpak does the same for AI agent capabilities. Learn why a registry matters.

Two Types of Capabilities

mpak distributes two types of agent capabilities:

MCP Bundles Tools and integrations that give agents the ability to do things, like query databases, call APIs, or interact with external systems.

Skills Knowledge and behaviors that make agents smarter at specific domains, like code review, legal analysis, or sales methodology.

The mpak Ecosystem

Component	URL	Purpose
Registry UI	mpak.dev	Browse and search capabilities
Registry API	registry.mpak.dev/docs	Programmatic access (OpenAPI)
CLI	@nimblebrain/mpak	Search, pull, run

Get Started

Quickstart Install the CLI and run your first MCP server in 2 minutes

Migrate Your MCP Server Publish an existing server to mpak in 10 minutes

Why a Registry? Why registries matter for MCP distribution

Browse the Registry Explore what's available on mpak.dev

External Resources

MCP Protocol - Model Context Protocol specification
MCPB Specification - Bundle format spec
Agent Skills Specification - Skills spec