# Agent Plugins

> An open, vendor-neutral standard for packaging reusable components—Agent Skills and MCP servers—into portable plugins that extend AI agents.

Agent Plugins is an open, vendor-neutral specification for packaging reusable components into portable plugins that extend AI agents. Version 1.0.0 is the current published release, developed publicly with a Technical Steering Committee that the project page states includes Core Maintainers from Amazon, Cursor, Microsoft, OpenAI, and Vercel.

## What It Is

Agent Plugins defines a minimal portable package format for two types of components: Agent Skills (following the Agent Skills specification) and MCP servers (following the Model Context Protocol specification). The goal is an interoperability floor that lets plugin authors write once and have compatible clients discover and load their plugins consistently, without duplicating or rearranging components for each client's proprietary format.

## The Package Structure

An Agent Plugin is a directory with a required `plugin.json` manifest and optional components in fixed locations:

- **`plugin.json`** — identifies the plugin and the Agent Plugins version it targets
- **`skills/`** — contains Agent Skills, each as a subdirectory with a `SKILL.md` file, optional scripts, and references
- **`mcp.json`** — describes stdio, Streamable HTTP, or legacy HTTP+SSE MCP servers
- **Reverse-domain extension namespaces** (e.g., `com.example.client/`) — let individual clients add behavior without changing the portable core

The smallest valid plugin is a directory with `plugin.json` and a single `SKILL.md` file, making the barrier to authorship very low.

## Open Development Model

The specification is openly licensed and developed in public on GitHub. Proposals and technical decisions are public, and participation is open to the broader ecosystem. Ideas for new features and material changes begin in GitHub Discussions, where proposals must establish a concrete portability need and implementer support before advancing. The repository includes a Technical Charter, governance documents, and future considerations files alongside the normative specification.

## Design Philosophy and Tradeoffs

Agent Plugins deliberately defines only a small interoperability floor. Distribution, installation, permissions, user experience, and client-specific capabilities remain entirely under each client's control. This narrow scope is intentional: it avoids prescribing how clients expose skills to users or models, while still giving plugin authors a predictable, shared structure for the portable parts. The result is a format that can coexist with client-specific extensions rather than replacing them.

## Update: Specification 1.0.0

The Agent Plugins Specification 1.0.0 is the current published release as of the repository's last update in August 2026. The GitHub repository was created in April 2026 and has accumulated 500 stars and 30 forks, signaling early ecosystem interest. JSON Schemas for both the plugin manifest and MCP configuration are published at versioned URLs under `agent-plugins.org/schemas/1.0.0/`.

## Features
- Portable plugin package format for AI agents
- Agent Skills packaging (SKILL.md-based)
- MCP server configuration (stdio, Streamable HTTP, HTTP+SSE)
- Required plugin.json manifest with JSON Schema validation
- Reverse-domain extension namespaces for client-specific behavior
- Vendor-neutral open standard
- Public governance and Technical Charter
- Compatible client discovery and loading
- Versioned JSON Schemas for plugin and MCP configuration
- Open GitHub Discussions for proposals

## Integrations
Agent Skills (agentskills.io), Model Context Protocol (MCP), Amazon, Cursor, Microsoft, OpenAI, Vercel

## Platforms
API, CLI, DEVELOPER_SDK

## Pricing
Open Source

## Version
1.0.0

## Links
- Website: https://agent-plugins.org
- Documentation: https://agent-plugins.org/specification
- Repository: https://github.com/agentplugins/agent-plugins-spec
- EveryDev.ai: https://www.everydev.ai/tools/agent-plugins