Workflow instance

Overview

This page shows options to configure and deploy a workflow, including parameter values and instance configuration.

Information on this page

• Workflow Name — The name of the workflow being deployed
• Parameters section — List of parameters defined in the workflow:Parameters with default values are pre-filled (can be modified) Required parameters without default values must be filled in before deployment Optional parameters can be left empty
• Instance Configuration — Opens the appropriate creation wizard to configure compute resources: either the Virtual Machine wizard or the Spot Instance wizard.

After resource configuration is set, displays CPU, RAM, Storage, and estimated price.

Actions you can perform on this page

• Add Compute Instance — Opens the VM creation page to configure CPU, RAM, storage, network, security groups, and SSH keys
• Add Spot Instance — Opens the spot creation page to configure CPU, RAM, storage, network, security groups, and SSH keys
• Create workflow instance — Button to start the workflow deployment; becomes active after VM is configured and all required parameters are filled

OpenClaw installation

OpenClaw is an open-source personal AI assistant that runs on your own devices. It connects to messaging channels (Telegram, WhatsApp, Slack, Discord, and more) through a WebSocket-based gateway. The assistant uses Claude Sonnet as the AI brain via the Anthropic API.

What You Get Out of the Box

• Claude Sonnet 4 as the AI brain (via Anthropic API)
• Telegram channel configured and ready
• Conversation memory and knowledge management
• Multi-channel support (add WhatsApp, Slack, Discord later)
• Auto-language detection (responds in whatever language you write)
• Extensible with skills and tools

Cost Estimate

Deployment Steps

Step 1: Create Telegram Bot

Create a new bot via @BotFather in Telegram.

1. Open Telegram, search for @BotFather
2. Send /newbot
3. Choose a display name (e.g. OpenClaw Assistant)
4. Choose a username ending in _bot (e.g. my_openclaw_bot)
5. Copy the token that looks like: 1234567890:AAH...

⚠️ Keep your bot token secret. Anyone with this token can control your bot.

Step 2: Get Your Chat ID

The bot needs your chat ID to know who to talk to.

1. Open your new bot in Telegram and send it /start
2. Open this URL in your browser (replace TOKEN with your bot token):

https://api.telegram.org/bot<TOKEN>/getUpdates

1. Find your chat ID in the JSON response: "chat":{"id":123456789,...}
2. Copy the numeric ID (e.g. 123456789)

Step 3: Get Anthropic API Key

Sign up at console.anthropic.com and create an API key.

Go to console.anthropic.com

1. Navigate to API Keys section
2. Click Create Key, copy the key starting with sk-ant-...
3. Add credits to your account ($5 minimum recommended)

Deploy an Instance

1. Go to Workflows, find OpenClaw AI Assistant, click Create Instance
2. Choose a VM (vCPU 2+, Memory 4GB+, Storage 16GB+)
3. Fill in OPENCLAW_TG_TOKEN, OPENCLAW_ANTHROPIC_KEY, OPENCLAW_CHAT_ID
4. Click Create workflow instance and wait 3–5 minutes
5. Check Telegram — your bot should greet you!

Start Chatting

Your bot is live. Just type any question in Telegram.

Try: ask anything in natural language, send /status for bot stats, or /help for available commands.

Deployment Script Reference

The script installs OpenClaw from npm, configures Telegram and Anthropic, and starts the service. Here is what it does step by step.

Execution Steps

The script runs as root on a fresh Ubuntu VM and executes 6 steps in sequence. If any step fails, the script aborts (set -euo pipefail).

Running the Script Manually

If emma.ms deployment failed or you want to run the script on your own Ubuntu VM, copy the script into a text editor(VS Code, Sublime, Notepad++, etc.) and save it as a file with a .sh extension(e.g., openclaw-vanilla.sh)

export OPENCLAW_TG_TOKEN="your-telegram-token" export OPENCLAW_ANTHROPIC_KEY="sk-ant-..." export OPENCLAW_CHAT_ID="123456789" export OPENCLAW_BOT_NAME="Kesha" export OPENCLAW_USER_NAME="YourName" chmod +x openclaw-vanilla.sh && sudo -E ./openclaw-vanilla.sh

The -E flag preserves environment variables when running with sudo.

After Deployment: SSH & Configuration

Once the VM is running, connect via SSH to customize your assistant.

Connecting via SSH

ssh -i <path to ssh key> <username>@<VM_IP>

Find your VM IP in the emma.ms dashboard under your instance details.

Key Files

Useful Commands

Troubleshooting

Bot didn’t send a message

• Check that OPENCLAW_CHAT_ID is correct (numbers only)
• Make sure you sent /start to the bot before deploying
• SSH in and check logs: journalctl -u openclaw --no-pager -n 50

Bot says API key is invalid

• Verify the Anthropic key at console.anthropic.com → API Keys
• Check that you have credits on your Anthropic account
• Verify config: cat /opt/openclaw/.openclaw/openclaw.json

emma.ms shows Error in callback

This is a known emma.ms platform issue. The VM may still have been created successfully.

• Check the emma.ms dashboard for your VM and its IP address
• SSH in and check: openclaw --version (if installed, the script ran)
• If not installed, run the script manually via SSH

Security Notes

• OpenClaw runs as a dedicated system user (openclaw), not root
• Config file is chmod 600 — only the openclaw user can read it
• Rotate your Telegram bot token if it was ever exposed
• Rotate your Anthropic API key periodically at console.anthropic.com