DevBridge AI Provider for Magnitude

DevBridge AI Provider for Magnitude connects WordPress to Magnitude, an open-source inference engine that runs open-weight models on your own hardware.

The plugin integrates Magnitude’s OpenAI-compatible API with the WordPress AI Client, allowing compatible WordPress plugins to use locally running AI models.

This is an independent project. It is not affiliated with, endorsed by, or sponsored by Magnitude AI Inc., the maker of Magnitude. The name Magnitude is used only to identify the service this plugin connects to.

Features:

• Text generation with models served by Magnitude.
• Tool (function) calling on supported models.
• Structured JSON output on supported models.
• Image and audio input when supported by the running model.
• Automatic model discovery.
• Model capabilities are detected from the capabilities reported by Magnitude.
• No API key required for local connections.
• Integration with the WordPress AI Client.
• Connector integration under Settings > Connectors.
• Settings page under Settings > Magnitude.
• Test Playground for trying prompts against running models.
• Adjustable temperature, top P, and maximum token settings in the Test Playground.
• Optional system instructions and example prompts.
• Ability to select a running model, stop an in-progress request, clear conversation history, and bypass the response cache.
• Connection status and server address information.

Requirements:

• WordPress 7.0 or higher.
• PHP 7.4 or higher.
• The Magnitude application installed and running.
• At least one model downloaded and started in Magnitude.

Magnitude lists models that are currently running. Start a model in the Magnitude application before expecting it to appear in the connector.

Important limitations:

• Magnitude provides text and chat inference. It does not provide image generation, embeddings, or speech generation through this connector.
• The default server URL is http://127.0.0.1:10100/inference.
• The first request to a model may take a minute while the model loads.
• Magnitude returns one answer per request, so generating multiple variations in a single request is not available through this plugin.
• Reasoning models may require a higher token limit to finish thinking and produce a complete answer.

External Services

This plugin connects to the Magnitude server configured in Settings > Magnitude or through the DEVBRIDGE_AI_PROVIDER_MAGNITUDE_HOST constant.

By default, the server address is http://127.0.0.1:10100/inference, which points to a Magnitude application running on the same computer as WordPress.

What data is sent:

• Prompts and context submitted by other plugins through the WordPress AI Client.
• Prompts submitted through the Test Playground.
• Requests to discover available models.
• The API key, if configured in Settings > Connectors.

When data is sent:

• When a plugin uses the AI Client with Magnitude.
• When an administrator submits a prompt through the Test Playground.
• When an administrator opens Settings > Connectors or Settings > Magnitude, or manually checks the connection.

The plugin does not send requests to Magnitude merely because it is activated, and it does not send data from the front end of the website on its own.

Where data is sent:

Data is sent only to the Magnitude server address configured by the site administrator. The plugin does not send prompts or usage data to the plugin author or any other third-party service.

Magnitude is a separate application developed by Magnitude AI Inc. For more information, visit https://magnitude.dev/.