name: azure-image-gen description: Generate a single image via Azure OpenAI Images REST API and return it inline in OpenClaw. homepage: https://learn.microsoft.com/azure/ai-services/openai/ metadata: openclaw: emoji: "🖼️" requires: bins: ["python3"] env: - AZURE_OPENAI_ENDPOINT - AZURE_OPENAI_API_KEY - AZURE_OPENAI_IMAGE_DEPLOYMENT - AZURE_OPENAI_API_VERSION primaryEnv: AZURE_OPENAI_API_KEY notes: | This skill uses Azure OpenAI REST (not OpenAI public API). It generates exactly ONE image per invocation and prints JSON to stdout: { "ok": true, "path": "/absolute/path/to/image.png" } OpenClaw should attach the file at path as an inline image response.

Azure Image Generation (OpenClaw)

This skill generates a single image using Azure OpenAI Images API and returns it for inline display in OpenClaw-supported channels.

How it works

• OpenClaw executes the script:

  python3 scripts/gen.py "<prompt>"
  

• The script:
  • Calls Azure OpenAI Images REST API
  • Saves the image using a timestamp-based filename
  • Prints a JSON object containing the absolute file path

Script location

skills/azure-image-gen/scripts/gen.py

Required environment variables

These are loaded from /home/thomas/.env.azure by the script:

• AZURE_OPENAI_ENDPOINT
• AZURE_OPENAI_API_KEY
• AZURE_OPENAI_IMAGE_DEPLOYMENT
• AZURE_OPENAI_API_VERSION

Example:

AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com
AZURE_OPENAI_API_KEY=****
AZURE_OPENAI_IMAGE_DEPLOYMENT=gpt-image-1.5
AZURE_OPENAI_API_VERSION=2025-04-01-preview

Output contract (important)

The script prints exactly one JSON object to stdout:

{
  "ok": true,
  "path": "/home/thomas/.openclaw/workspace/skills/azure-image-gen/out/20260206_101753.png",
  "prompt": "a cyberpunk cat in Hong Kong at night"
}

OpenClaw should treat path as a media attachment and send it inline.

Notes

• No batch generation
• No gallery / index.html
• No OpenAI public API usage
• Azure-only, REST-only