A Cross-Platform Catalog for AI Plugins, MCP Servers, and Agent Tools

You've probably felt it by now: every AI assistant has its own plugin ecosystem, its own naming conventions, and its own way of doing things. Codex plugins here, Claude Code skills there, MCP servers somewhere else entirely. Finding tools that actually work across the clients you use—without digging through a dozen scattered directories—is a genuine chore. Awesome AI Plugins is a community-maintained catalog that tries to solve exactly that, pulling plugins, skills, MCP servers, apps, and agent tools into one cross-platform list.

What It Does

At its core, this is a curated list. It collects extensions for a range of AI assistants—Codex, ChatGPT, Claude Code, Gemini CLI, Grok, Kimi, DeepSeek Harness, Cursor, OpenCode, and others—and organizes them into sections like Official Plugins, Community Plugins, and Formats & Development. Some listings target a single assistant; others support several or follow open standards such as Agent Skills and MCP.

The catalog is explicitly not a universal installer. It's a discovery layer. Each entry links to the original project, and you follow that project's own setup instructions for whatever clients and formats it supports. Alongside the human-readable README, the repo ships machine-readable compatibility exports in plugins.json and .agents/plugins/marketplace.json, which is a nice touch for anyone building registry or automation tooling on top of it. There's also a searchable HOL Plugin Registry if scrolling a long markdown file isn't your thing.

Why It's Cool

• It's honest about what it is. Plenty of "awesome" lists quietly pretend to be more than they are. This one says outright that it's a discovery catalog, not an installer, and that installation varies by client and project. That kind of clarity saves you from false expectations.

• Cross-platform by design. Rather than siloing tools by vendor, the catalog acknowledges that a single listing might work across multiple assistants or conform to open standards like MCP and Agent Skills. If you're juggling more than one AI client (and who isn't these days), that framing matters.

• Security is baked into the workflow. The recommended path includes validating your plugin with plugin-scanner and adding a scanner GitHub Action. Scanner CI is optional for listing, but projects that maintain it receive the full trust score, while those without it get a 10% reduction. It's a lightweight incentive structure that nudges the ecosystem toward continuous checking.

• Machine-readable exports. The plugins.json and marketplace JSON files mean this isn't just a list for humans. If you're building a registry, a CLI, or some automation that needs to know what's compatible with what, the data is already there in a parseable form.

• A clear contribution path. The Start Here section lays out a five-step workflow—choose your clients, build, validate, add scanner CI, ship—that gives new contributors a concrete on-ramp instead of vague encouragement.

How to Try It

This one's simple to explore, since it's a catalog rather than a library.

1. Head to the repository: github.com/hashgraph-online/awesome-ai-plugins
2. Browse the sections in the README, or use the searchable HOL Plugin Registry if you'd rather filter than scroll.
3. Pick a listing that matches your client, then follow that project's own installation guide.

If you're on the other side of things—building a plugin, skill, MCP server, or agent tool—the README recommends a local preflight before you submit:

pipx run plugin-scanner lint .
pipx run plugin-scanner verify .

From there, you can optionally add the HOL scanner GitHub Action for continuous security checks. The full details live in SCANNER_GUIDE.md and CONTRIBUTING.md. The project is Apache 2.0 licensed and welcomes pull requests.

Final Thoughts

Awesome AI Plugins is best understood as infrastructure for discovery, not a product you install and forget about. If you're a developer trying to keep tabs on what's available across multiple AI assistants, or someone building an extension and looking for a place to list it, this catalog gives you a reasonable starting point and a clear contribution path. The trust-score incentive and machine-readable exports show some thought went into keeping it useful beyond a static list. It won't install anything for you, and it won't vouch for every entry—but as a map of a fragmented landscape, it's a practical one to keep bookmarked.

Follow @githubprojects for more developer tools and open source projects.