🦀 OpenClaw deployed & generating waifus in under 5 minutes 🦀

Update: Check out this exciting announcement: - OpenClaw + Agentic use of Civitai = 🚀✨

tl:dr: Following on from my announcement of free OpenClaw hosting at Runware, I've put together this guide on how to actually use it; interacting with your agent via Discord, and generating images via the Civitai API.

🦀 Free OpenClaw - Recap

We've taken the hosting cost out of the equation. If you've considered playing with OpenClaw in the past, but have been put off by the config, or monthly cost of a managed host, we've got you covered. You don't pay Runware anything to run it for you - they manage storage, networking, updates, security, and you only pay for the tokens your agent (LLM) consumes.

There's no subscription, no minimum spend, no promotional pricing, ads, upsells, limits, freemium tier, etc. It's legitimately free hosting for your 'Claw agent. The only requirement is that you have credits in your account for LLM usage - but you don't even need those to deploy an instance and click around.

Prerequisites

Before you start, you’ll need a Runware account.

Note: you don't need to fund your Runware account to get started - you can deploy your instance to check things out, completely free of charge.

You’ll also want access to whichever messaging service you plan to connect. OpenClaw supports channels such as:

• Telegram

• Discord

• Slack

• WhatsApp

• Signal

• Google Chat

• iMessage

In this guide, we're going to cover configuring Discord.

🚀 Provisioning your Free Instance

Once you’re logged into Runware, launch the OpenClaw page and click Deploy in the top right corner.

Runware will begin provisioning your hosted instance. Behind the scenes, this creates an isolated OpenClaw container and wires it to your Runware account.

The deployment usually takes around one to three minutes.

Once the instance is ready, the Status will change to Running, and you’ll see an option to launch the OpenClaw Control UI under Actions.

Your OpenClaw Control UI is where the rest of the setup happens. From here, you can choose a model, configure channels, manage your agent, and start testing.

The important difference from a manual setup is that the LLM provider connection is already handled. You don’t need to generate a separate provider key, SSH into a server, or edit an environment file. Your hosted OpenClaw instance is already connected to Runware’s LLM model catalogue, giving you access to a range of LLMs - from frontier state-of-the-art models like Claude Opus 4.8, to balanced open-source options like the default, Qwen 3.5 397B.

⚙️ Enabling Channels

Now that your instance is deployed, we'll configure a Channel. Channels are how you interact with your Agent. You can always interact inside the OpenClaw Control UI, in the browser, but it's a far better experience to chat with your assistant via DMs in Discord, Telegram, or any of the other supported messaging apps.

Discord Setup

Discord setup is a little more involved than Telegram (widely considered to be the easiest), and requires a little finessing to help the bot along.

• The first step is to head to the Discord Developer Portal - https://discord.com/developers/applications - where you'll create a new App.

With that created, you'll need to do some config, and gather some information;

• From the General Information page, you need to note the Application ID.

• From the Bot page, you'll need to click the Reset Token button, and grab the generated secret token.

• From the Bot page, you'll need to enable all three Privileged Gateway Intent toggles.

• From the OAuth 2 page, you'll check Bot under OAuth2 URL Generator

• Then from the same page, you'll check Administrator under Bot Permissions.

Note: while this is the fastest way to get a Discord bot up and running - it's advisable to allocate sensible permissions to your agent, especially if it's going to be used by multiple users.

• Make sure the Installation Type is Guild Install, and copy the URL to clipboard.

• Paste the URL into a browser, and you'll be prompted, via Discord, to add the bot to any servers you have Admin access to. I highly recommend setting up a personal server to deploy your agent to.

At this stage, it's normal for the bot to show up as Offline.

• There are a couple of pieces of information we need from Discord to proceed, and you will have to enable Discord Developer mode from the User Settings > Developer > Developer Mode toggle.

• You'll need to note down the Server ID for the Server your agent has been invited to. Right click on the Server icon > Copy Server Info > Copy Server ID.

• You'll also need the Channel ID, if you want to interact with your agent in a particular Channel of the server. Right click on a channel > Copy Channel ID.

Now we can jump back into the OpenClaw Control UI.

• First tell the agent;

Install the Discord plugin: openclaw plugins install @openclaw/discord
Add "discord" to plugins.allow list
Add discord entry to plugins.entries
Restart the gateway

Then

Configure Discord channel with
Bot token: <your bot token>
Application ID: <your application id>
Server ID: <your server id>
Channel ID: <your channel id>

You may have to offer some encouragement, but eventually the agent will report Success.

• If the agent reports success, but the Discord bot appears offline, paste this link - https://github.com/openclaw/openclaw/issues/31260 - to the agent in the OpenClaw Control UI, explaining that it contains a common bug/solution.

The agent will make some changes to the configuration, and your Discord bot should come online.

Your first interaction with the Discord bot will be a pairing string - paste the code into the Control UI;

And you're now free to interact, without the need to log into the OpenClaw Control console.

🖌️ Making Stuff

You can use your new agent to interact with the APIs & SDKs of any service. There are Skills - pre-packaged "instructions" which help OpenClaw understand how to perform tasks - but it can also build its own scripts and knowledge from information provided.

Civitai has an official OpenClaw Skill in development which will make querying the API and generating content a breeze. For now, we're going to chat with The Great God Om, our image gen agent, (via Discord) to get set up.

First, create an API key from your Civitai account settings.

Note, it's a good idea to set a sensible limit when generating your API key! This prevents accidental Buzz spend above your intended budget.

Then tell your agent something like;

"Save this API key in a .env file in my workspace, then write a script that uses the Civitai API to generate images. Here's the Civitai API Documentation."

It may take some time, but eventually it'll report success. If it doesn't, chat with it a bit more, give it specific models, and specific API Recipes - https://developer.civitai.com/orchestration/recipes/

And from there, you can try some free-text prompting;

😎 Tips & Tricks

• Give your agent a personality - they'll add it to their memory, and answer in character.

• Set up recurring (Cron) jobs for scheduled tasks, from the Cron Jobs menu item in OpenClaw Control.

❓ Troubleshooting

One thing you'll notice is that the agent asks for Exec Approval for many actions. This is fine when working within the OpenClaw Control UI, but when you're in Discord it doesn't route those to you, by default.

To enable Exec Approvals in Discord, head to the OpenClaw Control UI > Settings > Channels > Discord > Expand Exec Approvals > Make sure it's Enabled.

Then you'll see;

✨ The Future

There’s plenty more we can do from here. OpenClaw supports more channels, more advanced agent configuration, custom skills, memory, and a lot of practical workflows.

In a future guide we’ll look at more useful OpenClaw use cases, deeper configuration, and how to start shaping your agent into a useful collaborator.

If you try the hosted setup on Runware, let me know what you build, what breaks, and what you’d like to see improved!