Skip to content

Railway Deployment

Railway provides managed cloud hosting with GitHub integration. Deploy ClawBio with a few clicks — no Dockerfile required.

1. Connect Your Repository

  1. Sign up at railway.app and link your GitHub account
  2. Click New Project → Deploy from GitHub Repo
  3. Select your ClawBio fork or the ClawBio/ClawBio repository

Railway auto-detects Python projects and installs requirements.txt dependencies.

2. Set Environment Variables

In the Railway dashboard, go to Variables and add:

Variable Value
LLM_API_KEY Your LLM provider API key
LLM_BASE_URL Provider endpoint (optional — defaults to OpenAI)
CLAWBIO_MODEL Model name (e.g. gemini-2.0-flash)
TELEGRAM_BOT_TOKEN From @BotFather
RATE_LIMIT_PER_HOUR 10 (or your preferred limit)

Use Google Gemini for free hosting

Pair Railway's free tier with Gemini's free API (1,500 req/day) for a zero-cost bot deployment.

3. Configure the Start Command

In Settings → Deploy, set the start command:

python3 bot/roboterri.py

Or for Discord:

python3 bot/roboterri_discord.py

4. Persistent Storage

By default, Railway containers are ephemeral — files written during a session are lost on redeploy. For persistent skill output:

  1. Go to New → Database → Volume
  2. Mount the volume at /app/results
  3. Configure skills to write output to /app/results

This keeps genomic reports and analysis results across deploys.

5. Custom Domains

  1. Go to Settings → Networking → Custom Domain
  2. Add your domain (e.g. bot.clawbio.ai)
  3. Add the CNAME record Railway provides to your DNS

6. Monitoring

Railway provides built-in logging and metrics:

  • Logs: Real-time stdout/stderr from your bot
  • Metrics: CPU, memory, and network usage
  • Deploy history: Rollback to any previous deployment

Cost Considerations

Plan Included Overage
Hobby $5/mo, 8 GB RAM, 8 vCPU $0.000231/min
Pro $20/mo, more resources Usage-based

A Telegram bot running ClawBio skills typically uses minimal resources — the LLM provider handles the heavy computation. The main cost is your LLM API usage, not Railway hosting.

Example: Full Stack

Deploy the bot with a local Ollama instance for fully self-hosted LLM:

Railway project
├── clawbio (GitHub deploy)
│   ├── Start: python3 bot/roboterri.py
│   └── Env: LLM_BASE_URL=http://ollama:11434/v1
└── ollama (Docker image: ollama/ollama)
    └── Volume: /root/.ollama

GPU not available on Railway

Railway does not currently offer GPU instances. For ML-heavy skills (scRNA embedding, structure prediction), use Docker with GPU passthrough or a private cloud deployment.

Next: Private Cloud — self-hosted deployment for sensitive data.