Railway MCP Server
Official# `@railway/mcp-server` is deprecated
Railway MCP is now bundled into the Railway CLI.
This npm package no longer contains the standalone TypeScript MCP server. It is a compatibility shim that launches:
```bash
railway mcp
```
## Migration
Install or upgrade the Railway CLI:
```bash
bash <(curl -fsSL https://railway.com/install.sh)
```
Then configure supported MCP clients:
```bash
railway mcp install
```
To configure the hosted MCP server instead of the local stdio server:
```bash
railway mcp install --remote
```
You can also configure a local MCP client directly with:
```json
{
"mcpServers": {
"railway": {
"command": "railway",
"args": ["mcp"]
}
}
}
```
Remove old client entries that run:
```bash
npx -y @railway/mcp-server
```
## Compatibility
For existing configs that still invoke `@railway/mcp-server`, this package delegates to `railway mcp`.
If the Railway CLI is missing, the package exits with migration instructions.
## Docs
See the Railway CLI MCP docs: https://docs.railway.com/cli/mcp
TDQS
Scored across 14 tools
Most tools have distinct purposes targeting specific Railway resources like projects, environments, services, deployments, and variables, with clear action verbs. However, 'deploy' and 'deploy-template' could be confused as both handle deployment, though their descriptions clarify one is for directory uploads and the other for templates.
Tool names predominantly follow a consistent verb-noun pattern (e.g., create-environment, list-services, set-variables), with clear and readable conventions. Minor deviations include 'check-railway-status' using a verb-noun-adjective structure and hyphenated names throughout, but overall consistency is high.
With 14 tools, the count is well-scoped for managing Railway projects, covering setup, deployment, logging, and resource management. Each tool serves a clear purpose without redundancy, fitting typical server tool ranges and the domain's complexity.
The tool set provides comprehensive coverage for Railway operations, including project and environment creation, deployment, logging, variable management, and listing resources. There are no obvious gaps; agents can perform full lifecycle management from setup to monitoring.