How to Create Thumbnails and Graphics with OpenClaw

If you publish content — YouTube videos, blog posts, social media — thumbnails and graphics are a recurring task. They’re not complex, but they take time and usually require a separate tool. When OpenClaw has image generation available through AgentPatch, you can request thumbnails and graphics directly through your Telegram or Discord chat.

Why This Matters

Thumbnails have specific requirements: they need to read well at small sizes, convey the topic quickly, and match a consistent visual style if you publish regularly. Describing those requirements to an AI image generator and getting usable results is something Recraft handles reasonably well, especially for illustration-style output.

AgentPatch includes Recraft image generation in its tool marketplace. Connect AgentPatch to OpenClaw and you can generate thumbnails and graphics on demand without a Recraft account or any additional configuration.

Setup

The AgentPatch CLI is designed for AI agents to use via shell access. Install it, and your agent can discover and invoke any tool on the marketplace.

Install (zero dependencies, Python 3.10+):

pip install agentpatch

Set your API key:

export AGENTPATCH_API_KEY=your_api_key

Example commands your agent will use:

ap search "web search"
ap run google-search --input '{"query": "test"}'

Get your API key from the AgentPatch dashboard.

Install the AgentPatch skill from ClawHub — it teaches OpenClaw when to use AgentPatch and how to use the CLI:

clawhub install agentpatch

MCP Server (Alternative)

If you prefer raw MCP tool access instead of the skill, add AgentPatch to ~/.openclaw/openclaw.json:

{
  "mcp": {
    "servers": {
      "agentpatch": {
        "transport": "streamable-http",
        "url": "https://agentpatch.ai/mcp",
        "headers": {
          "Authorization": "Bearer YOUR_API_KEY"
        }
      }
    }
  }
}

Replace YOUR_API_KEY with your actual key from the AgentPatch dashboard. Restart OpenClaw and it discovers all AgentPatch tools automatically.

Example

You’re preparing a YouTube video about setting up a home server. You message your OpenClaw bot on Telegram:

“Generate a YouTube thumbnail for a video called ‘Build Your Own Home Server in 2026’. It should look professional, have a tech/hardware feel, and work at small sizes. Bold visual, minimal text in the image itself.”

OpenClaw calls the Recraft tool through AgentPatch and sends back the thumbnail image in your chat.

If you want to try a different direction:

“Try a version that’s more abstract — a server rack illustration with a glowing blue light, dark background, clean and modern.”

OpenClaw generates a second version. You compare and pick one.

For social media graphics:

“Create a square graphic for an Instagram post announcing a new blog post called ‘Five Terminal Tools I Use Every Day’. Simple design, developer aesthetic, readable at small sizes.”

OpenClaw calls Recraft through AgentPatch with the new parameters and delivers the image to your chat.

You can also use context from other tools in the same request:

“Search for the top trending topics in the dev community this week and generate a thumbnail concept for a video about the most interesting one.”

OpenClaw calls Google Trends or Google News through AgentPatch to find the topic, then calls Recraft to generate a matching thumbnail.

Wrapping Up

OpenClaw with AgentPatch turns your messaging app into a lightweight design tool for thumbnails and graphics. The same connection also covers search, news, email, maps, and more. Browse everything at agentpatch.ai.

← Back to blog