← Back to blog

npm MCP Package Tutorial: Install & Configure Conjra for AI Editor Infrastructure

If you're using Claude Code, Cursor, or Windsurf and want to provision cloud infrastructure without leaving your editor, you need an npm MCP package that bridges the gap. Conjra is exactly that — an MCP-native npm package that gives your AI editor direct access to 18+ cloud providers.

This tutorial covers installing and configuring the conjra npm MCP package, connecting providers, and using it to provision real infrastructure.

Step 1: Install the npm MCP Package

Conjra is published on npm and installs as a global CLI:

npm install -g conjra

Once installed, verify it works:

conjra --version

Step 2: Initialize for Your AI Editor

Conjra works with 20 AI editors. To register it as an MCP server, run:

conjra init --ai all

This configures every supported editor. You can also target a specific one:

conjra init --ai claude    # Claude Code only
conjra init --ai cursor    # Cursor only
conjra init --ai windsurf  # Windsurf only

Running conjra init with no flag auto-detects which editors you have installed. This registers Conjra's MCP server in your editor's config file — your editor now has access to all of Conjra's infrastructure tools.

Step 3: Connect a Cloud Provider

Conjra stores credentials in an encrypted local vault. Connect your first provider:

conjra add supabase
conjra add stripe
conjra add railway

You'll be prompted for API keys. These are encrypted with AES-256-GCM and stored locally — they're never sent anywhere except directly to the provider's API.

Step 4: Provision Infrastructure from Your Editor

With Conjra configured, you can now provision infrastructure using natural language. In your AI editor, simply ask:

"Use conjra to create a Supabase project called my-app"
"Use conjra to create a Stripe product called Pro Plan at $29/month"
"Use conjra to deploy this project to Railway"

Conjra handles the API call, returns the result, and your AI editor incorporates the response into your workflow. No switching contexts, no copying credentials, no clicking through dashboards.

Why Use an npm MCP Package?

Installing an MCP npm package like conjra is the fastest way to add infrastructure provisioning to your AI editor workflow. Unlike SaaS alternatives that route your API calls through third-party middleware, Conjra is a native MCP package that calls provider APIs directly from your machine. Your credentials stay local, there's no vendor lock-in, and you can use it offline.

The MCP ecosystem is growing rapidly. By installing an npm MCP package today, you're adopting a standard that works across Claude Code, Cursor, Windsurf, and the 16+ other editors Conjra supports.

Install conjra — the npm MCP package for cloud infrastructure.