OpenClaw¶

In this article

• Openclaw. Core Features
• Deployment Features
• Getting Started After Deploying Openclaw
• Connection and Initial Settings
• Connecting an AI chat bot on your own server or LLM model from the marketplace
• Interacting with OpenClaw via Command Line
• Connecting to the Web Interface
• Connecting a Telegram Bot
• Ordering a Server with OpenClaw via API

Openclaw. Core Features¶

• Email and calendar management
• Executing terminal commands and working with files
• Browser automation (search, form filling, data extraction)
• LLM integration: Claude, GPT, local models via Ollama (OpenClaw is free, but requires costs for API keys of the used AI models)
• Background scheduled work (heartbeat loop: e.g., checking tasks every 30 minutes) without constant user supervision
• Extensibility: supports "skills" (plugins) that can be created by yourself or taken from the community

OpenClaw is a powerful tool for developers and advanced users who want to create a personal AI assistant with full data control. However, due to its broad access rights and the project's youth, its use requires technical literacy and caution in security configuration.

Deployment Features¶

• Installation time 15-30 minutes including OS;
• Minimum requirements: Node.js 22+, 2 GB RAM.

Getting Started After Deploying Openclaw¶

After paying for the order, a notification of the server's readiness to work will be sent to the email address specified during registration. It will contain the VPS IP address, as well as the login and password for connection. Our company's clients manage the hardware via the server management panel and API — Invapi.

Connection and Initial Settings¶

After obtaining access to the server, you need to establish a connection to it via SSH with superuser (root) rights:

ssh root@<server_ip>

sudo su - openclaw
openclaw onboard --install-daemon

openclaw onboard

Follow these steps:

1. Agree with the risk warning – select Yes.
2. At the Onboarding mode step, select QuickStart.

3. At the next step Model/auth provider:

   • choose Skip for now if you plan to connect our AI agents;
   • if you are connecting an external AI agent – select it and on the next step enter the API key

   

4. At the Filter models by provider step, select All providers.

5. At the Default model step, set it to Keep current.
6. The next step allows you to set up the management channel via messengers; we suggest returning to this question later and selecting Skip for now.
7. We also suggest not configuring skills immediately: Configure skills now? – select No.

8. To skip configuring automatic user commands at the Enable hooks? step, press Space and Enter.

   

9. At the step How do you want to hatch your bot? we choose Do this later

At this point the initial setup is complete. Openclaw will show you text with useful tips for further configuration and a message of successful installation.

Connecting an AI chat bot on your own server or LLM model from the marketplace¶

You can connect the desired neural network model to OpenClaw not only via external APIs, but also by deploying it on your own server. This will give you greater control over resources and increase data privacy.

To order a personal chat bot with an integrated model, use the following instructions: this guide.

We are ready to offer you various LLM models – you can always choose and order the necessary one in the section of our website Language LLM Models.

So, the server is ordered; to connect it to OpenClaw follow the instructions below:

1. To connect our agent, run the command:

   curl -fsSL https://raw.githubusercontent.com/hkadm/ollama_gpu_test/refs/heads/main/scripts/hostkey_openclaw.sh -o ./hostkey_openclaw.sh
   

2. Make the file executable and run:

   chmod +x ./hostkey_openclaw.sh
   ./hostkey_openclaw.sh
   

3. The script HOSTKEY AI CHATBOT Agent Provider Setup will be launched. On the Chatbot Domain step specify the domain name of your server (For example: https://ollama<Server_ID_from_Invapi>.hostkey.in).

4. On the API Key from OpenWebUI step enter the API key obtained on your server with OpenWebUI (Go to OpenWebUI in the Settings menu - Account - the key will be in the Api Key field. If there is no key, generate a new one.)

5. On the Model selection step choose the required model from the available ones (you can enter the model number from the list).

6. Set the Context parameter to a value between 8000 and the maximum supported by your model. The Max Tokens parameter can be left at the default.

Done – you can now chat with OpenClaw using the command line.

Interacting with OpenClaw via Command Line¶

To manage via the command line, use the following command to launch:

openclaw tui

To exit the command-line management mode, press Ctrl + C twice.

Connecting to the Web Interface¶

The web interface is a convenient option for managing Clawbot. It is intended for control and debugging; through it you can chat with the bot, manage tokens, and monitor the agent’s status.

The web interface is not available for use on the open internet; it can only be used on a local computer. To provide access, several configuration steps are required:

1. Connect to the server with the agent via SSH and run the command:

   openclaw dashboard
   

2. Note the output of the command. For further configuration, we need the values from the 1^st and 4^th lines:

   

3. On your local computer, open the command line again and enter the command:

   ssh -N -L 18789:127.0.0.1:18789 openclaw@IP_of_server
   

   You may need to re-enter the password to access the server. Press Enter – Clawbot will create a tunnel for use on the local machine.

4. In the browser window, enter the Dashboard URL – the web interface will open and you can start using it.

Connecting a Telegram Bot¶

The main convenience of managing OpenClaw is undoubtedly the ability to manage and monitor using messengers. Consider an example of connecting via Telegram. For connecting to other messengers, use the same algorithm, which will differ only in the procedure of creating a bot in the specific messenger.

1. Using the built‑in tool @BotFather, create a bot to chat with OpenClaw:

   • find @BotFather
   • enter the command /newbot
   • enter the bot name, which will appear in your channel list
   • enter the bot username (must end with “bot”)

2. When the bot is created, it will be assigned an API key, which you will use for integration.

   

3. Restart the OpenClaw setup wizard with the command:

   openclaw onboard
   

4. At the Select channel step, choose Telegram (Bot API). Enter the API key.

5. Go to Telegram, find and launch the created bot. The bot will provide a Pairing code for pairing with the agent.

   

6. In Clawbot settings, finish pairing. Enter the command:

   openclaw pairing approve telegram <Pairing Code>
   

Now you can manage the agent using the Telegram bot.

Ordering a Server with OpenClaw via API¶

To install this software using the API, follow these instructions.