Deployment

ZeptoClaw can be deployed anywhere a Linux binary can run. Choose the method that fits your infrastructure.

One-click deploy

Any VPS

The fastest way to deploy ZeptoClaw is using the automated setup script:

curl -fsSL https://zeptoclaw.com/setup.sh | bash

This interactive wizard will:

• Download the latest ZeptoClaw binary
• Guide you through configuring your LLM provider (Anthropic or OpenAI)
• Set up your messaging channel (Telegram, Slack, Discord, or Webhook)
• Install and start a systemd service

Docker deployment:

curl -fsSL https://zeptoclaw.com/setup.sh | bash -s -- --docker

Uninstall:

curl -fsSL https://zeptoclaw.com/setup.sh | bash -s -- --uninstall

Docker (single container)

The simplest production deployment:

FROM ghcr.io/qhkm/zeptoclaw:latest

COPY config.json /root/.zeptoclaw/config.json

EXPOSE 8080
CMD ["zeptoclaw", "gateway"]

docker build -t my-agent .
docker run -d --name zeptoclaw \
  -e ZEPTOCLAW_PROVIDERS_ANTHROPIC_API_KEY=sk-ant-... \
  -e ZEPTOCLAW_CHANNELS_TELEGRAM_BOT_TOKEN=123456:ABC... \
  my-agent

Docker Compose

For multi-service setups:

version: '3.8'
services:
  zeptoclaw:
    image: ghcr.io/qhkm/zeptoclaw:latest
    restart: unless-stopped
    environment:
      - ZEPTOCLAW_PROVIDERS_ANTHROPIC_API_KEY=${ANTHROPIC_KEY}
      - ZEPTOCLAW_CHANNELS_TELEGRAM_BOT_TOKEN=${TELEGRAM_TOKEN}
    volumes:
      - zeptoclaw-data:/root/.zeptoclaw
    healthcheck:
      test: ["CMD", "zeptoclaw", "config", "check"]
      interval: 30s
      retries: 3

volumes:
  zeptoclaw-data:

Fly.io

Deploy to Fly.io with a single command:

fly.toml

app = "my-zeptoclaw"
primary_region = "sin"

[build]
  image = "ghcr.io/qhkm/zeptoclaw:latest"

[env]
  RUST_LOG = "info"

[[services]]
  internal_port = 8080
  protocol = "tcp"

  [[services.ports]]
    port = 443
    handlers = ["tls", "http"]

fly launch
fly secrets set ZEPTOCLAW_PROVIDERS_ANTHROPIC_API_KEY=sk-ant-...
fly secrets set ZEPTOCLAW_CHANNELS_TELEGRAM_BOT_TOKEN=123456:ABC...
fly deploy

Railway

Deploy via Railway CLI:

railway init
railway up

# Set secrets
railway variables set ZEPTOCLAW_PROVIDERS_ANTHROPIC_API_KEY=sk-ant-...
railway variables set ZEPTOCLAW_CHANNELS_TELEGRAM_BOT_TOKEN=123456:ABC...

Render

Use the Render dashboard or render.yaml:

services:
  - type: worker
    name: zeptoclaw
    runtime: docker
    dockerfilePath: ./Dockerfile
    envVars:
      - key: ZEPTOCLAW_PROVIDERS_ANTHROPIC_API_KEY
        sync: false
      - key: ZEPTOCLAW_CHANNELS_TELEGRAM_BOT_TOKEN
        sync: false

Systemd (bare metal)

For direct deployment on a Linux server:

/etc/systemd/system/zeptoclaw.service

[Unit]
Description=ZeptoClaw AI Agent
After=network-online.target

[Service]
Type=simple
User=zeptoclaw
ExecStart=/usr/local/bin/zeptoclaw gateway
Restart=always
RestartSec=5
Environment=RUST_LOG=info
EnvironmentFile=/etc/zeptoclaw/env

[Install]
WantedBy=multi-user.target

sudo systemctl enable zeptoclaw
sudo systemctl start zeptoclaw

Health checks

ZeptoClaw’s config check command can be used as a health check:

zeptoclaw config check

Returns exit code 0 if configuration is valid.

Persistent data

Important directories to persist across restarts:

Path	Contents
~/.zeptoclaw/config.json	Configuration
~/.zeptoclaw/memory/	Long-term memory
~/.zeptoclaw/sessions/	Conversation history
~/.zeptoclaw/skills/	Custom skills
~/.zeptoclaw/plugins/	Custom plugins

Mount ~/.zeptoclaw as a volume in Docker deployments.

Resource requirements

ZeptoClaw is very lightweight:

Resource	Requirement
CPU	Any modern CPU (ARM64 or x86_64)
RAM	~6MB RSS at idle
Disk	~4MB binary + data
Network	Outbound HTTPS to LLM provider APIs