Most MCP servers connect to AI clients the same way: install Node.js, edit a local config file, manage credentials by hand, and repeat the whole thing every time you switch machines or onboard a new client. MCP Express skips all of that. You configure your server through a UI and connect your AI client via OAuth — a browser redirect, a login, and you're done.
This guide walks you through the full setup using Claude as the example, but the same OAuth approach works for any MCP-compatible AI client — including Cursor and GitHub Copilot.
What OAuth Actually Changes
The CLI approach ties your MCP connection to a specific machine. The config lives locally, so your setup on a work laptop and your home machine are two separate configurations — and web-based AI clients aren't an option at all.
With OAuth, your MCP server lives in MCP Express, not on your machine. That means:
- Works with web and desktop AI clients — not just desktop apps. Use Claude.ai, Cursor, GitHub Copilot, or any MCP-compatible client from any browser, on any machine.
- One setup per server, not per machine — connect once, access everywhere. No re-configuration when you switch environments.
- Revoke access in one place — manage, restrict, or disconnect any client's access from your MCP Express dashboard at any time.
Set Up and Connect in 3 Steps
By the end of this guide, your AI client will have direct, controlled access to your MCP Express server — and through it, any tools you've configured (databases, APIs, cloud services, or custom integrations). The steps below use Claude, but the server setup is identical regardless of which AI client you connect.
Once connected, you can ask your AI things like:
- "Summarise the last 10 rows added to the orders table."
- "Check CloudWatch for any errors in the past hour on the production stack."
- "Draft a client update based on the latest entries in the project database."
The connection is scoped entirely to what you expose — your AI client can't reach anything outside your configured tools.
Prerequisites
Before you begin, make sure you have:
Estimated time: Under 5 minutes.
Step 1: Create Your MCP Server
In your dashboard, navigate to MCP Servers → New MCP Server.
Name it based on the context — keeping names like "Client-A-PostgreSQL" or "Internal-DevOps" makes it easier to manage multiple servers across client projects.
Click Create MCP Server.
Click View Details on your newly created server. This is where you control what your AI client can and can't do.
From here you can activate integrations, store credentials, and monitor activity logs. Credentials are encrypted via AWS KMS — the same standard used in banking infrastructure. You can rotate credentials without touching your server configuration or triggering a redeployment.
Supported integrations include:
- Databases: PostgreSQL, MySQL, MSSQL, DynamoDB
- Cloud: AWS CloudWatch, S3
- Communication: Slack, Confluence, Ghost CMS
- Infrastructure: SSH, Kubernetes
- Other: REST APIs, External MCP Servers
Full integrations list can be found in our documentation.
Step 3: Connect Claude via OAuth
This is where OAuth removes the manual setup entirely. In Claude, navigate to Customize → Connectors and click the + icon.
Give the connector a name and set the URL to:
https://api.mcp-express.com/gateway/mcp
One connector per MCP server
Each MCP Express server requires its own connector in Claude. If you have multiple servers, you'll need to repeat this step for each one. To keep things organised, use your MCP server name as the connector name — so "Client-A-PostgreSQL" in MCP Express becomes "Client-A-PostgreSQL" in Claude. Much easier to manage when you're running several at once.
Claude will redirect you to MCP Express to complete the OAuth flow. Log in, grant access, and select the MCP server you just configured. No further configuration required.
You can manage, restrict, or revoke access at any time from your MCP Express dashboard — without touching any config files.
Alternative: Manual Connection via CLI
If you prefer the manual configuration route, you can connect your MCP server via the CLI instead. It takes under 5 minutes and covers the same end result — use whichever fits your setup.
Your MCP Server Is Connected — Start Querying
Open your AI client and ask a question in plain English. It uses your configured tools to interact with your infrastructure and returns a structured answer.
Connect your next integration or add more tools to your MCP server in seconds from the same dashboard.
Try it Now — Free tier, no credit card required. Your first MCP server takes under 5 minutes to set up.
Further Resources:
- Documentation — Every supported integration, configuration option, and code example in one place.
- Contact Us — Got a question before signing up, or want to talk through your setup? Drop us an email.
- Open a Support Ticket — Already inside the app and something's not working? Open a ticket directly from your dashboard.
