MCP Server Bundles Hub: Install 500+ Integrations

Overview

MCP Bundles Hub is a curated catalog of provider integrations and deployment bundles designed for MCP (Model Context Protocol) servers. The repository aggregates 500+ ready-to-use manifests and metadata entries so developers and platform operators can discover, deploy, and manage integrations for models, providers, connectors, and orchestration components from a single catalog.

Running a Bundles Hub gives an “app store” experience for MCP: browse available provider bundles, download their manifests, and apply them to your MCP server or deployment pipeline. The hub can be hosted publicly or privately and consumed by an MCP server or tooling that accepts bundle manifests and URLs, speeding up onboarding and standardizing integrations across environments.

Features

• Catalog of 500+ provider integrations and bundles (manifests, icons, metadata)
• Standardized bundle manifests describing provider, version, and install artifacts
• Simple static hosting model (serve JSON/YAML manifests over HTTP)
• Easily consumable by MCP servers or CI pipelines via URL or downloaded manifest
• Searchable metadata and tags (provider, capability, version)
• Scripts and tooling for validating and previewing bundles (where included)
• Support for private catalogs or local development instances

Installation / Configuration

Prerequisites:

• A running MCP server or management tooling that accepts bundle manifests or bundle URLs
• Git and a static HTTP server (Docker + nginx, Python http.server, etc.)
• Optional: curl, jq for automation

Clone the Bundles Hub repository:

git clone https://github.com/thinkchainai/mcpbundles.git
cd mcpbundles

Option A — Serve the hub as a static HTTP catalog (simple):

# From the repository root, serve the 'hub' or 'bundles' directory over HTTP
# Example: use Python to serve port 8080
cd hub
python3 -m http.server 8080
# Hub will be available at http://localhost:8080/

Option B — Serve with Docker + nginx (production-like):

# Serve the hub directory via nginx container on port 8080
docker run --rm -p 8080:80 \
  -v "$(pwd)/hub:/usr/share/nginx/html:ro" \
  nginx:stable-alpine

Configure your MCP server to consume the catalog URL. The exact configuration key depends on your implementation; the general idea is to point the MCP server at the hub index URL:

# Example MCP server configuration snippet (adapt to your MCP server)
bundlesCatalog:
  name: "internal-bundles"
  url: "https://catalog.example.com/index.json"
  refreshInterval: "5m"

Or use an environment variable (example):

export MCP_BUNDLES_CATALOG_URL="https://catalog.example.com/index.json"

After configuring, restart or instruct your MCP server to refresh catalogs so the new bundles appear in discovery.

Available Resources

The repository organizes bundle content and tooling to make discovery and automation straightforward.

Bundle manifest fields commonly include: id, name, version, provider, description, tags, iconUrl, manifestUrl (link to the install artifact).

Use Cases

1. Install a provider bundle into a running MCP server (via URL)
   • Host the hub and obtain the bundle manifest URL, e.g.: https://catalog.example.com/bundles/openai-provider-1.0.json
   • In MCP tooling that accepts bundle URLs, point it at the manifest URL or use a POST API to install:

     curl -X POST "https://mcp.example/api/v1/bundles" \
       -H "Content-Type: application/json" \
       -d '{"bundleUrl":"https