Deploy and Host OpenClaw on Railway

Source: https://github.com/lefant/openclaw-railway-template

Just like clawdbot-railway-template but with support for OpenAI Codex (ChatGPT Plus) authenticated GPT-5 models via device code OAuth flow.

About Hosting

OpenClaw is an AI coding assistant platform that runs as a self-hosted gateway. This template wraps OpenClaw with a web-based setup wizard, automatic reverse proxy, and persistent storage on Railway. Authentication with OpenAI is handled through a browser-based device code flow — no API keys needed if you have a ChatGPT Plus subscription.

Why Deploy

• No API key required — sign in with your ChatGPT account using device code auth (like Codex CLI)
• One-click Railway deployment — setup wizard guides you through configuration in the browser
• Persistent state — config, credentials, and workspace survive redeploys via Railway Volume
• Multi-provider support — works with Anthropic, OpenAI, Google, and custom OpenAI-compatible providers
• Built-in management — debug console, config editor, backup export/import, health checks

Common Use Cases

• Running an AI coding assistant accessible from anywhere via web UI
• Hosting a Telegram/Discord/Slack bot powered by GPT-5 or Claude
• Self-hosting an AI gateway with OAuth authentication instead of managing API keys
• Setting up a shared AI assistant for a team with persistent workspace

Dependencies for

Deployment Dependencies

• Railway Volume mounted at /data for persistent config and workspace storage
• Public networking enabled (assigns a *.up.railway.app domain)
• SETUP_PASSWORD environment variable (protects the setup wizard)
• Optional: OPENCLAW_GATEWAY_TOKEN for stable gateway auth across redeploys
• Optional: OPENCLAW_VERSION to pin a specific OpenClaw release tag