---
title: AI
description: Integrate Railway with AI assistants using plugins, connectors, Agent Skills, and the Railway MCP Server.
---
Railway provides tools for integrating with AI coding assistants, letting you manage your infrastructure through natural language.

## Integration options

### Railway Agent

The [Railway Agent](/ai/railway-agent) is a chat-based assistant built directly into the Railway dashboard. It can manage services, diagnose failing deployments, and open pull requests to fix issues in your code.

- Runs inside the Railway dashboard with no setup required
- Creates and configures services, variables, databases, and networking
- Automatically diagnoses failed deployments and proposes fixes

[Get started with the Railway Agent →](/ai/railway-agent)

### Agent Connectors

[Agent Connectors](/ai/agent-connectors) give the Railway Agent access to the tools you already use. Connect Notion, Linear, or Sentry, or point Railway at your own remote MCP server, then attach a connector as context in dashboard chat or dev.new.

- Connect once in workspace settings, attach per conversation
- Bring your own MCP server over OAuth or a static header
- Connectors are personal, so credentials are never shared with the workspace

[Get started with Agent Connectors →](/ai/agent-connectors)

### Cloud agents

[Cloud agents](/cloud-agents) run Claude Code, Codex, or Grok CLI on a persistent Railway virtual machine, signed in as you. Launch one from the terminal and reconnect to it later with your work still in place.

- Runs the coding agent you already use, with your own credentials
- Keeps its disk between runs, and sleeps when you disconnect
- Manages agents and sessions from `railway ca`

[Get started with cloud agents →](/cloud-agents)

### Agent integrations

[Agent integrations](/ai/agent-integrations) bring the Railway Agent into your team chat on Slack and Discord. Mention **@Railway** to inspect deployments, read logs, and make changes without leaving the conversation.

- Install from your account settings with one click per platform
- Each teammate acts with their own Railway access; usage bills to the linked workspace
- Currently in beta

[Get started with agent integrations →](/ai/agent-integrations)

### Agent skills

[Agent Skills](/ai/agent-skills) are an open format for extending AI coding assistants with specialized knowledge about Railway. The `use-railway` skill guides AI agents to perform tasks like deploying services, managing environments, and querying metrics.

- Works with Claude Code, Cursor, OpenAI Codex, and OpenCode
- Install with a single command
- Covers project setup, deployments, networking, observability, and more

[Get started with Agent Skills →](/ai/agent-skills)

### Plugins and connectors

Railway provides [official plugins and connectors](/ai/plugins-and-connectors) for ChatGPT, Codex, Claude, Claude Code, Grok Build, and Cursor. These integrations package Railway workflows for each assistant and connect to the hosted Railway MCP server.

- Install from the plugin or connector directory built into your assistant
- Use the `use-railway` skill and hosted Railway MCP server from supported plugin hosts
- Connect Claude directly through Railway OAuth without a local CLI installation

[Get started with plugins and connectors →](/ai/plugins-and-connectors)

### MCP server

The [Railway MCP Server](/ai/mcp-server) implements the Model Context Protocol, enabling direct communication between AI assistants and your Railway infrastructure. Connect to the server at `mcp.railway.com` through the CLI, which reuses your `railway login` credentials without a second authentication.

- One command to install: `railway mcp install`
- Add `--oauth` for direct OAuth
- Create projects, deploy templates, manage environments, pull variables
- Works with Cursor, VS Code, Claude Code, Codex, Copilot, Factory Droid, OpenCode, Windsurf, Cline, and Devin

[Get started with MCP Server →](/ai/mcp-server)