Awesome Plugins

An open marketplace that bundles reusable plugins and skills for AI agents, providing manifests, discovery, and installation flows that work across multiple agent platforms.

awesome-plugins is a multi-agent plugin marketplace maintained by Licentora. Each plugin in this repository ships per-agent manifests so the same plugin can be installed natively in every supported AI coding agent — no rewriting, no agent-specific forks.

Supported agents

Agent	Root marketplace manifest	Per-plugin manifest
Claude Code	`.claude-plugin/marketplace.json`	`plugins/<name>/.claude-plugin/plugin.json`
Codex	`.agents-plugin/marketplace.json`	`plugins/<name>/.codex-plugin/plugin.json`
Cursor	`.cursor-plugin/marketplace.json`	`plugins/<name>/.cursor-plugin/plugin.json`

A plugin only needs the manifests for the agents it actually targets — there is no requirement to support all three. Skills and other components live alongside the manifests under the plugin's folder.

Installation

The repository as a whole is registered as one marketplace per agent. After you register it once, every plugin inside plugins/ becomes available to install or browse through your agent's own commands.

The canonical source for all three agents is:

licentora/awesome-plugins
# or, when a full URL is required:
https://github.com/licentora/awesome-plugins

Claude Code

Register the marketplace (one-time setup):

claude plugin marketplace add licentora/awesome-plugins

Pin a specific release by appending a ref:

claude plugin marketplace add licentora/awesome-plugins@v1.0.0

Browse available plugins:
```
claude plugin marketplace list
```
Install a plugin by name:
```
claude plugin install <plugin-name>
```
If a plugin name is ambiguous across registered marketplaces, qualify it:
```
claude plugin install awesome-plugins/<plugin-name>
```
Verify the install:
```
claude plugin list
```

If claude is not available in your shell (e.g., Claude Desktop without the CLI), add the marketplace from the Plugins UI inside Claude Desktop using the same source string.

Codex

Codex uses a two-step flow: register the marketplace from your shell, then install plugins from inside the Codex CLI. codex plugin install is not available as a shell command — the install flow lives inside the Codex CLI's interactive /plugins command.

Register the marketplace from your shell:

codex plugin marketplace add licentora/awesome-plugins

Install plugins from inside the Codex CLI:
- Open the Codex CLI.
- Run /plugins.
- Select the awesome-plugins marketplace.
- Pick which plugins to install from the list shown.

If codex is not available in your shell (e.g., Codex Desktop only), open Codex Desktop, navigate to Plugins / Plugin Directory, and paste the source licentora/awesome-plugins.

Cursor

Cursor manages plugins through an interactive agent REPL, not a single shell command.

From a shell inside your project, open the Cursor agent CLI:
```
agent
```

Inside the agent REPL, register the marketplace:

/plugin marketplace add https://github.com/licentora/awesome-plugins

Browse the available plugins:
```
/plugin list
```
Follow Cursor's prompts to enable a plugin you find. If you are unsure which plugin subcommands your Cursor version exposes, run /help inside the agent REPL.

Team / enterprise admins can alternatively use the Cursor Dashboard:

Dashboard -> Settings -> Plugins -> Team Marketplaces -> Import
Paste: https://github.com/licentora/awesome-plugins

Browsing available plugins

The fastest way to see what is offered is to ask your agent after registering the marketplace:

Agent	Command
Claude Code	`claude plugin marketplace list`
Codex	`/plugins` inside the Codex CLI (browse and install there)
Cursor	`/plugin list` inside the `agent` REPL

To browse without registering, look directly at the plugins/ directory of this repository on GitHub. Each subfolder is one plugin with its own README.md, skills, and manifests.

Repository structure

awesome-plugins/
├── .agents-plugin/
│   └── marketplace.json          # Codex marketplace manifest
├── .claude-plugin/
│   └── marketplace.json          # Claude Code marketplace manifest
├── .cursor-plugin/
│   └── marketplace.json          # Cursor marketplace manifest
├── plugins/
│   └── <plugin-name>/
│       ├── .claude-plugin/
│       │   └── plugin.json       # Claude Code plugin manifest
│       ├── .codex-plugin/
│       │   └── plugin.json       # Codex plugin manifest
│       ├── .cursor-plugin/
│       │   └── plugin.json       # Cursor plugin manifest
│       ├── skills/
│       │   └── <skill-name>/
│       │       └── SKILL.md      # one skill per folder
│       ├── agents/               # (Claude Code, optional) custom sub-agent definitions
│       ├── hooks/                # (Claude Code, optional) hooks.json for event handlers
│       ├── monitors/             # (Claude Code, optional) monitors.json for background watchers
│       ├── bin/                  # (Claude Code, optional) executables added to PATH
│       ├── .mcp.json             # (Claude Code, optional) MCP server configuration
│       ├── .lsp.json             # (Claude Code, optional) LSP server configuration
│       ├── settings.json         # (Claude Code, optional) default settings on enable
│       └── README.md             # per-plugin docs
├── package.json
├── LICENSE
└── README.md

Source of truth. Each plugin folder under plugins/ is self-contained. The root marketplace.json files only index what is offered to each agent — the actual code, manifests, and skills live under the plugin folder. Adding a plugin to the marketplace means (a) creating its folder under plugins/ and (b) listing it in the relevant root marketplace manifest.

Optional directories. Only .<agent>-plugin/plugin.json is required for each agent you support. Everything else (agents/, hooks/, monitors/, bin/, .mcp.json, .lsp.json, settings.json) is opt-in based on what your plugin actually provides. These optional components follow the Claude Code plugin spec; Codex and Cursor support different (and smaller) sets of plugin components — consult each agent's plugin docs for what is honored.

Adding a new plugin

To contribute a plugin to this marketplace:

1. Scaffold the plugin folder

mkdir -p plugins/<plugin-name>/skills/<skill-name>

2. Add per-agent plugin manifests

Add one manifest for each agent you want to support. Required fields vary slightly per agent; check the schema references below for the full list.

Claude Code — plugins/<plugin-name>/.claude-plugin/plugin.json:

{
  "$schema": "https://www.schemastore.org/claude-code-plugin-manifest.json",
  "name": "<plugin-name>",
  "version": "1.0.0",
  "description": "What the plugin does.",
  "author": { "name": "Your Name", "email": "you@example.com" },
  "license": "MIT"
}

Codex — plugins/<plugin-name>/.codex-plugin/plugin.json:

{
  "name": "<plugin-name>",
  "version": "1.0.0",
  "description": "What the plugin does.",
  "skills": "./skills/",
  "interface": {
    "displayName": "<Display Name>",
    "category": "Developer Tools"
  }
}

Cursor — plugins/<plugin-name>/.cursor-plugin/plugin.json:

{
  "name": "<plugin-name>"
}

3. Add at least one skill

plugins/<plugin-name>/skills/<skill-name>/SKILL.md:

---
name: <skill-name>
description: What the skill does and when to use it.
allowed-tools: Read Grep Glob
---

# Skill body — instructions Claude follows when the skill is invoked.

For multi-agent plugins, prefer a single combined skill with shared discovery rules plus per-agent workflow sections (Claude shell, Codex hybrid, Cursor REPL). See plugins/find-plugins/skills/find-plugins/SKILL.md for the reference layout. If the agents' workflows diverge too much to share one file, split into per-agent skills named <plugin-name>-<agent> instead.

4. Register the plugin in the marketplace manifests

Claude Code — append to plugins[] in .claude-plugin/marketplace.json:

{
  "name": "<plugin-name>",
  "description": "What the plugin does.",
  "author": { "name": "Your Name" },
  "source": "./plugins/<plugin-name>",
  "category": "developer-tools",
  "homepage": "https://github.com/licentora/awesome-plugins"
}

Codex — append to plugins[] in .agents-plugin/marketplace.json:

{
  "name": "<plugin-name>",
  "source": {
    "source": "local",
    "path": "./plugins/<plugin-name>"
  },
  "policy": {
    "installation": "AVAILABLE",
    "authentication": "ON_INSTALL"
  },
  "category": "Developer Tools"
}

5. Add a per-plugin README

Write plugins/<plugin-name>/README.md documenting what the plugin does, which agents it supports, and how to use it after install.

6. Open a pull request

Include in the PR description: which agents the plugin targets, a short demo or screenshot, and how a reviewer can sanity-check the skill locally.

Schema references

When authoring or validating manifests, point your editor at these schemas for autocomplete and live validation:

Manifest	Schema
Claude Code marketplace	`https://www.schemastore.org/claude-code-marketplace.json`
Claude Code plugin	`https://www.schemastore.org/claude-code-plugin-manifest.json`
Cursor plugin reference	https://cursor.com/docs/plugins

Codex manifest schemas are tracked through the Codex project; consult the Codex CLI documentation for the current authoritative reference.

Trust and safety

Installing a plugin grants it the tools and permissions declared in its manifests and skills. Before installing any plugin from this — or any — marketplace:

Inspect the manifest. Check allowed-tools declarations in each skill, any hooks/, bin/, .mcp.json, and .lsp.json for what the plugin actually does at runtime.
Inspect the source. Each plugin in this repository lives under plugins/<name>/ — read the code before enabling.
Pin a ref for production. Use licentora/awesome-plugins@<tag> (or the equivalent in your agent) so future commits cannot silently change behavior.
Report concerns. Open an issue at https://github.com/licentora/awesome-plugins/issues if a plugin's behavior does not match its documentation.

Plugins shipped from this marketplace follow the safety patterns documented in each plugin's skills — explicit user confirmation before destructive actions, no silent installs of additional sources, and clear handling of failure modes.

License

MIT — see LICENSE.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Awesome Plugins

Table of contents

Supported agents

Installation

Claude Code

Codex

Cursor

Browsing available plugins

Repository structure

Adding a new plugin

1. Scaffold the plugin folder

2. Add per-agent plugin manifests

3. Add at least one skill

4. Register the plugin in the marketplace manifests

5. Add a per-plugin README

6. Open a pull request

Schema references

Trust and safety

License

About

Uh oh!

Contributors

Uh oh!

Languages

Name		Name	Last commit message	Last commit date
Latest commit History 15 Commits
.agents-plugin		.agents-plugin
.claude-plugin		.claude-plugin
.cursor-plugin		.cursor-plugin
.github		.github
plugins		plugins
.gitignore		.gitignore
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
CONTRIBUTING.md		CONTRIBUTING.md
LICENSE		LICENSE
README.md		README.md
SECURITY.md		SECURITY.md
package-lock.json		package-lock.json
package.json		package.json

Folders and files

Latest commit

History

Repository files navigation

Awesome Plugins

Table of contents

Supported agents

Installation

Claude Code

Codex

Cursor

Browsing available plugins

Repository structure

Adding a new plugin

1. Scaffold the plugin folder

2. Add per-agent plugin manifests

3. Add at least one skill

4. Register the plugin in the marketplace manifests

5. Add a per-plugin README

6. Open a pull request

Schema references

Trust and safety

License

About

Topics

Resources

License

Code of conduct

Contributing

Security policy

Uh oh!

Stars

Watchers

Forks

Contributors

Uh oh!

Languages