← Blog Guides & tutorials • 4 min read Connect Claude, Cursor, VS Code and Windsurf to your deploy platform with MCP
Follow this on your own server. Start your 14-day trial
Every AI editor now speaks the Model Context Protocol, but each one keeps its server list in a different place, in a slightly different format. This guide has the exact configuration for the five clients people actually use, pointed at Railyard’s MCP server. The patterns carry over to any remote MCP server that supports OAuth.
The endpoint is the same everywhere:
https://app.railyard.run/api/mcp
It’s a remote server over Streamable HTTP. Every client below signs in through your browser with OAuth, so no token goes in a config file unless you want one there.
Claude (web, desktop and mobile)
- Open Customize → Connectors.
- Search for Railyard and choose Connect.
- A Railyard page opens. Pick the team and the access level (
read,deployorwrite), then approve.
The connector is then available in every Claude surface on your account. On Team and Enterprise plans, an organization owner adds the connector first; members then connect with their own Railyard accounts, each capped at their own Railyard role.
Claude Code
claude mcp add --transport http railyard https://app.railyard.run/api/mcp
Inside Claude Code, run /mcp, pick railyard and choose Authenticate. Your browser opens the consent page, and Claude Code catches the callback on a local loopback port.
The server is added for your user by default. To share it with everyone working in a repository, add --scope project. That writes a .mcp.json at the repo root with only the URL, so it’s safe to commit. Each person still signs in with their own account.
Cursor
Add the server to .cursor/mcp.json in your project, or ~/.cursor/mcp.json for all projects:
{
"mcpServers": {
"railyard": { "url": "https://app.railyard.run/api/mcp" }
}
}
Open the MCP section of Cursor’s settings. Railyard appears with a sign-in prompt. Complete it in the browser and the tools load.
VS Code (Copilot agent mode)
VS Code uses a different shape: a top-level servers key and an explicit transport type. Create .vscode/mcp.json:
{
"servers": {
"railyard": {
"type": "http",
"url": "https://app.railyard.run/api/mcp"
}
}
}
Don’t leave out "type": "http". Without it, VS Code assumes a local stdio server and tries to run the URL as a command. Start the server from the CodeLens above the entry, or from the MCP view. VS Code opens the browser sign-in on first use.
Windsurf
Windsurf keeps its list in ~/.codeium/windsurf/mcp_config.json (on Windows, %USERPROFILE%\.codeium\windsurf\mcp_config.json). Remote servers use serverUrl:
{
"mcpServers": {
"railyard": { "serverUrl": "https://app.railyard.run/api/mcp" }
}
}
Refresh the MCP servers in Cascade’s panel, then sign in when prompted.
Using an API token instead
Some clients, scripts and CI jobs can’t do a browser sign-in. For those, create a token under Settings → Access & tokens: choose its scope and set an expiry. Then send it as a header.
Claude Code:
claude mcp add --transport http railyard https://app.railyard.run/api/mcp \
--header "Authorization: Bearer rly_..."
Cursor:
{ "mcpServers": { "railyard": { "url": "https://app.railyard.run/api/mcp",
"headers": { "Authorization": "Bearer rly_..." } } } }
VS Code, reading the token from a prompted input so it isn’t stored in the file:
{
"inputs": [{ "type": "promptString", "id": "railyard-token", "description": "Railyard API token", "password": true }],
"servers": {
"railyard": {
"type": "http",
"url": "https://app.railyard.run/api/mcp",
"headers": { "Authorization": "Bearer ${input:railyard-token}" }
}
}
}
Windsurf, reading it from an environment variable:
{ "mcpServers": { "railyard": { "serverUrl": "https://app.railyard.run/api/mcp",
"headers": { "Authorization": "Bearer ${env:RAILYARD_TOKEN}" } } } }
Tokens follow the same rules as OAuth connections. Their scope never goes beyond the role of the person who created them, and every call is in the activity log.
Check the connection
Whatever the client, ask:
> Which Railyard team am I connected to, and what access does this connection have?
The agent calls whoami and answers with the team, the plan and the effective access level. If that level is lower than you picked, your role on that team caps it. The authorization post explains why.
Then a read-only smoke test:
> List my apps and their status.
Which access level to pick
| If the agent should… | Pick |
|---|---|
| Answer questions about apps, deploys, logs and traffic | read |
| Also deploy, roll back, restart, scale and toggle maintenance | deploy |
| Also create apps, change variables and domains, run commands, query data | write |
Start with read in clients you use for general coding, and keep write for the one session where you actually manage infrastructure. You can connect the same client twice under different names with different scopes.
Troubleshooting
- The client says the server returned 405. Something is sending a GET. Make sure the client is set up for Streamable HTTP (
"type": "http"in VS Code), not the old SSE transport. - The sign-in loops. Remove the server, add it again, and sign in once. If the token was revoked under Settings → Access & tokens, a fresh sign-in is required.
- Tools are missing. Most clients cache the tool list. Restart the MCP server from the client’s MCP panel.
- “This API token can’t …” The connection’s access level is too low for that tool. The message names the level it needs.
Once connected, try deploying from Claude Code or debugging a failed deploy. Every tool is listed in the MCP docs.
Why people pick Railyard for this
- Your own cloud account. Deploy to DigitalOcean, Hetzner, Vultr, AWS, OVHcloud, or any Linux server over SSH, billed directly by your provider with no markup.
- One flat price per team. Launch is $22 a month with 2 servers and 1 seat. Scale is $45 with 5 servers and 5 seats, Team is $79 with 10 servers and 15 seats, and Agency is $149 with 20 servers and unlimited seats. Unlimited apps on every plan, no charge per app, per deploy or per gigabyte, and a 14-day trial to start.
- One-click templates. Deploy open-source apps, like databases, analytics and dev tools, to your own server in one click.
- Managed databases. Postgres, MySQL and Redis with scheduled, restore-verified backups and point-in-time recovery.
- Push-to-deploy. Deploy from GitHub, GitLab or the CLI, with automatic HTTPS and zero-downtime releases.