# Railway

> Spin up Railway projects, services, and deploys via remote MCP with OAuth.

[Canonical HTML page](https://top-mcps.com/mcp/railway) · [server.json](https://top-mcps.com/mcp/railway.json) · [methodology](https://top-mcps.com/about/methodology)

## Install

### Claude Desktop — `claude_desktop_config.json`

Paste under mcpServers. Fully quit and reopen Claude after editing.

```json
{
  "mcpServers": {
    "railway": {
      "command": "claude",
      "args": [
        "mcp",
        "add",
        "--transport",
        "http",
        "railway",
        "https://mcp.railway.com"
      ]
    }
  }
}
```

### Claude Code — `CLI or .mcp.json`

Run from your repo. Commit .mcp.json to share with your team.

```shell
claude mcp add railway -- claude mcp add --transport http railway https://mcp.railway.com
```

### Cursor — `.cursor/mcp.json`

Global path: ~/.cursor/mcp.json. Reload window after editing.

```json
{
  "mcpServers": {
    "railway": {
      "command": "claude",
      "args": [
        "mcp",
        "add",
        "--transport",
        "http",
        "railway",
        "https://mcp.railway.com"
      ]
    }
  }
}
```

### VS Code — `.vscode/mcp.json`

VS Code uses the "servers" key (not "mcpServers").

```jsonc
{
  "servers": {
    "railway": {
      "command": "claude",
      "args": [
        "mcp",
        "add",
        "--transport",
        "http",
        "railway",
        "https://mcp.railway.com"
      ]
    }
  }
}
```

### Windsurf — `~/.codeium/windsurf/mcp_config.json`

Open via Cascade → hammer icon → Configure.

```json
{
  "mcpServers": {
    "railway": {
      "command": "claude",
      "args": [
        "mcp",
        "add",
        "--transport",
        "http",
        "railway",
        "https://mcp.railway.com"
      ]
    }
  }
}
```

### Cline — `cline_mcp_settings.json`

Open via the Cline sidebar → MCP Servers → Edit.

```json
{
  "mcpServers": {
    "railway": {
      "command": "claude",
      "args": [
        "mcp",
        "add",
        "--transport",
        "http",
        "railway",
        "https://mcp.railway.com"
      ]
    }
  }
}
```

### Continue — `~/.continue/config.json`

Continue uses modelContextProtocolServers with a transport block.

```json
{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "stdio",
          "command": "claude",
          "args": [
            "mcp",
            "add",
            "--transport",
            "http",
            "railway",
            "https://mcp.railway.com"
          ]
        }
      }
    ]
  }
}
```

### Codex CLI — `~/.codex/config.toml`

Codex uses TOML. Each server is a [mcp_servers.<name>] subtable.

```shell
# ~/.codex/config.toml
[mcp_servers.railway]
command = "claude"
args = [
  "mcp",
  "add",
  "--transport",
  "http",
  "railway",
  "https://mcp.railway.com",
]
```

### Zed — `~/.config/zed/settings.json`

Zed calls them "context_servers". Settings live-reload on save.

```jsonc
{
  "context_servers": {
    "railway": {
      "command": {
        "path": "claude",
        "args": [
          "mcp",
          "add",
          "--transport",
          "http",
          "railway",
          "https://mcp.railway.com"
        ]
      }
    }
  }
}
```

### ChatGPT — `ChatGPT → Settings → Connectors (remote URL)`

Paste this into Settings → Connectors → Add custom connector. Requires ChatGPT Pro / Team / Enterprise / Edu with Developer mode enabled. ChatGPT supports remote HTTPS MCP servers only — stdio servers must be hosted on a public HTTPS endpoint first.

```json
{
  "name": "Railway",
  "transport": "http",
  "url": "https://mcp.railway.com"
}
```

## At a glance

- **Maintainer:** Railway
- **Transport:** Streamable HTTP, stdio
- **Auth model:** OAuth 2.1
- **Required secrets:** None
- **Supported clients:** Claude, Cursor, Any MCP-compatible client, Railway accounts
- **Latest version:** latest
- **Last verified:** 2026-05-27
- **Score:** 54/100 (rubric 2026-04 — see https://top-mcps.com/about/methodology)
- **Homepage:** https://docs.railway.com/ai/remote-mcp-server

## Quick answer

**What it does.** Surfaces Railway's control plane as MCP tools over a hosted HTTP endpoint: projects, services, deployments, environment variables, and a delegated agent tool.

**Best for:**
- Zero-config remote MCP setup
- OAuth-based team installs
- Quick project + service creation
- Triggering redeploys from chat
- Delegating multi-step ops to Railway's agent

**Not for:**
- Workflows that require the full local CLI surface
- Self-hosted MCP needs (this one is hosted)
- Bulk environment management

## Description

The official Railway MCP is a hosted remote server at mcp.railway.com. Agents authenticate via browser OAuth — no local install — and can create projects, deploy services, redeploy, and hand multi-step ops to Railway's own AI agent.

## Why it matters

Railway's remote MCP is one of the cleanest hosted MCP experiences in 2026 — OAuth, no JSON config, and a hand-off to Railway's own AI for harder tasks.

## Key features

- Hosted at mcp.railway.com — no install
- OAuth in browser
- Project + service + deploy tools
- `railway-agent` tool for multi-step ops
- Local stdio variant via `railway setup agent`

## FAQ

### Do I need to install anything?

No — the remote MCP is hosted at mcp.railway.com. Browser OAuth covers authentication.

### What's the `railway-agent` tool?

A tool that hands a natural-language request to Railway's own AI for multi-step operations like provisioning and deploying together.

### Is `accept-deploy` reversible?

No — it commits staged changes. Railway prompts for confirmation before invoking it.

## Changelog

- **2026-05-27** — Refreshed install snippets and fact sheet; verified for 2026.
- **2025-09-01** — Initial directory listing.