Model Context Protocol

Talk to HyperStore from any AI.

HyperStore MCP turns the world's first AI apps directory into a native tool that Claude, ChatGPT, Cursor, Windsurf, Cline, Zed and Gemini can use directly — no plugins, no scraping, no API keys.

Install in one line
$ uvx hyperstore-mcp
Add to Claude View on GitHub

What is MCP, in 30 seconds?

The Model Context Protocol is an open standard from Anthropic that lets any AI assistant call external tools the same way a developer calls an API — but in plain conversation. Think of it as USB-C for AI: one protocol, every client, every data source.

Native tool calls

Your AI doesn't browse the web or guess. It calls our directory directly and gets structured, up-to-date answers.

No API key required

Read endpoints are public. Install, connect, ask. Nothing to register, nothing to bill.

Works everywhere

Streamable HTTP, SSE and stdio transports — every major MCP client is supported out of the box.

What HyperStore MCP can do

Eight first-class tools your assistant can call on your behalf.

search_apps

Full-text search across every app — name, description, features.

ai_search

Natural-language semantic search: "a writing assistant cheaper than Jasper".

get_app

Full details for a single app: pricing, features, pros/cons, social links.

list_apps

Paginated listing with filters (pricing model, has API, open source, rating).

list_categories

Every curated category with live app counts.

category_apps

All apps in a category, sorted by popularity or rating.

browse_apps

Editorially curated picks — trending, new, staff favorites.

get_homepage

The full homepage in one shot — featured apps, top categories, hero copy.

Install in your AI of choice

Pick your client. Pick your transport. Copy. Done.

Claude Desktop

claude.ai/download

Open Settings → Developer → Edit Config, then add:

{
  "mcpServers": {
    "hyperstore": {
      "url": "https://mcp.store.hypergpt.ai/mcp"
    }
  }
}

Claude Code

CLI

One command from your terminal:

claude mcp add --transport http \
  hyperstore https://mcp.store.hypergpt.ai/mcp

ChatGPT Desktop

Connectors

In Settings → Connectors → Add a custom connector, paste:

https://mcp.store.hypergpt.ai/mcp

Cursor

cursor.com

Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "hyperstore": {
      "url": "https://mcp.store.hypergpt.ai/mcp"
    }
  }
}

Windsurf

windsurf.com

Open Cascade → MCP Servers → Add Custom:

{
  "mcpServers": {
    "hyperstore": {
      "serverUrl": "https://mcp.store.hypergpt.ai/mcp"
    }
  }
}

Zed

zed.dev

In your Zed settings under context_servers:

"context_servers": {
  "hyperstore": {
    "source": "custom",
    "url": "https://mcp.store.hypergpt.ai/mcp"
  }
}

Cline

VS Code extension

Open the MCP panel → Configure MCP Servers:

{
  "mcpServers": {
    "hyperstore": {
      "transportType": "streamableHttp",
      "url": "https://mcp.store.hypergpt.ai/mcp"
    }
  }
}

Gemini CLI

Google AI

From your terminal:

gemini mcp add --transport http \
  hyperstore https://mcp.store.hypergpt.ai/mcp

Universal — any MCP client

Local install via uvx

Add this entry to any MCP-compatible client's config (Claude Desktop, Cursor, Windsurf, Cline, Zed, etc):

{
  "mcpServers": {
    "hyperstore": {
      "command": "uvx",
      "args": ["hyperstore-mcp"]
    }
  }
}

Requires uv. The package runs locally and proxies to store.hypergpt.ai — no data leaves your machine other than the directory queries themselves.

Find HyperStore across the MCP ecosystem

We publish to every major MCP registry so your client can discover us automatically.

mcp.so Public registry Smithery Install in one click Glama MCP server catalog Pulse MCP Curated directory Awesome MCP GitHub list PyPI uvx hyperstore-mcp GHCR Docker image MCP Spec modelcontextprotocol.io

Questions teams ask us

Do I need an API key?

No. Read tools (search, list, get) are open and unauthenticated. We rate-limit per IP.

Can I self-host the MCP server?

Yes — the source is MIT-licensed on GitHub. Clone, run with uvx, point your client at it.

Will the directory stay free?

The public read API powering MCP is free and stays free. Write operations (listing your app) follow the normal submission flow.

How do I discover the server automatically?

We publish a standard discovery document at /.well-known/mcp.json. Clients that support auto-discovery will find us with just the domain. View discovery document →

Ready to plug your AI into the world's largest AI apps directory?

Install in under 30 seconds. No signup, no key, no friction.

Install now Read the docs