mcp -aoai-dalle3

This MCP server exposes an Azure OpenAI DALL-E 3 integration via MCP. It can generate images using the generate_image tool by sending a text prompt and optional parameters such as size, quality, and style. After an image is generated, you can use the download_image tool to retrieve and store the resulting image locally by providing the image URL, a local directory path, and a desired filename. To operate this server, ensure your Azure OpenAI settings (endpoint, API key, and deployment name) are configured in the environment variables. The server relies on the standard MCP workflow: clients request image generation through generate_image, receive a response containing image details or a URL, and optionally download the image to local storage with download_image. The default deployment name for DALL-E 3 is dalle3, but you can override it with AZURE_OPENAI_DEPLOYMENT_NAME if your resource uses a different deployment.

Prerequisites:

• Node.js 14+ and npm installed
• Access to an Azure OpenAI resource with DALL-E 3 deployment
• Internet access to install dependencies and fetch packages

Steps:

1. Clone or prepare the MCP server repository for mcp-server-aoai-dalle3.
2. Install dependencies:

npm install

3. Configure environment variables (see Environment Variables in README):

export AZURE_OPENAI_ENDPOINT="https://<your-resource>.openai.azure.com/"
export AZURE_OPENAI_API_KEY="<your-api-key>"
export AZURE_OPENAI_DEPLOYMENT_NAME="dalle3"  # optional, default if omitted
export OPENAI_API_VERSION="2024-02-15-preview"  # optional, default if omitted

4. Build the server:

npm run build

5. Run the server (as part of the MCP setup, typically invoked by your MCP orchestrator):

npm start

Notes and tips:

• Ensure AZURE_OPENAI_ENDPOINT and AZURE_OPENAI_API_KEY are kept secure and not embedded in client-side code.
• If you use a deployment name other than dalle3, set AZURE_OPENAI_DEPLOYMENT_NAME accordingly.
• The OPENAI_API_VERSION should align with what your Azure OpenAI resource supports; verify compatibility with your deployment.
• The download_image tool requires a valid image URL and write permissions to the specified localPath.
• When upgrading the server or dependencies, re-build to ensure the MCP client can load the latest build/index.js.
• If you encounter authentication errors, double-check that the API key has access to the Azure OpenAI resource and that the endpoint URL is correct.