ms-foundry-image-gen（MS Foundry 图像生成器）

Azure Foundry Image Generation

AI image generation using an Azure Foundry (Cognitive Services / OpenAI) images deployment. Returns raw image bytes (PNG/JPEG) or a URL depending on the deployment response.

Overview

• Requires network access to your Foundry endpoint and a valid API key.

Usage

Set environment variables (example):

export FOUNDRY_ENDPOINT="https://aif-sbxe2e-ai-agent-02.cognitiveservices.azure.com/"
export FOUNDRY_API_KEY="<your_api_key>"
export FOUNDRY_DEPLOYMENT="FLUX-1.1-pro"
export FOUNDRY_API_VERSION="2025-04-01-preview"

Generate an image (safe example using jq to build JSON):

# Basic validation (reject obviously malformed endpoints)
if ! printf '%s' "${FOUNDRY_ENDPOINT:-}" | grep -Eq '^https?://[A-Za-z0-9._:-]+/?$'; then
  echo "FOUNDRY_ENDPOINT looks unsafe or is not set" >&2
  exit 1
fi

url="${FOUNDRY_ENDPOINT%/}/openai/deployments/${FOUNDRY_DEPLOYMENT}/images/generations?api-version=${FOUNDRY_API_VERSION:-2025-04-01-preview}"

PROMPT="a red fox"
jq -n --arg prompt "$PROMPT" '{prompt:$prompt, n:1, size:"1024x1024", output_format:"png"}' | \
  curl --fail --show-error --silent \
    --url "$url" \
    -H 'Content-Type: application/json' \
    -H "api-key: ${FOUNDRY_API_KEY}" \
    --data-binary @- -o /tmp/generation_result.json
  
# Stream base64 payload to avoid storing large values in shell variables
jq -r '.data[0].b64_json' /tmp/generation_result.json | base64 --decode > /tmp/generated_image.png
echo "Image saved to: /tmp/generated_image.png"

Options

• FOUNDRY_ENDPOINT (required): Azure base URI for Foundry (include scheme, e.g. https://<name>.cognitiveservices.azure.com/)
• FOUNDRY_API_KEY (required): API key (primary credential)
• FOUNDRY_DEPLOYMENT (required): Deployment name to call
• FOUNDRY_API_VERSION (optional): API version (default: 2025-04-01-preview)

Notes

• The skill manifest (src/manifest.json) declares the required environment variables and marks FOUNDRY_API_KEY as the primary credential.
• This document provides a safe example using jq --arg and streaming to prevent shell interpolation and command-injection risks.

Troubleshooting

• If you see authentication errors, verify FOUNDRY_API_KEY permissions for the deployment.
• If jq or base64 are missing, install them via your package manager (e.g., apt install jq coreutils on Debian/Ubuntu).

License / Attribution

This skill is a minimal wrapper around the Foundry images generation REST endpoint for use in OpenClaw workflows.