WP Manifestindependent plugin directory
manifest / ai / ai-provider-for-magnitude

AI Provider for Magnitude

WordPress AI Client provider connector for Magnitude: run open-weight models locally.

by Mahbub Hussain · github.com/devmahbubhussain/ai-provider-for-magnitude · website

★ 0stars
0forks

Install

No release zip yet. The repository archive installs, but the folder name will carry the branch suffix and updates will not flow:

wp plugin install https://github.com/devmahbubhussain/ai-provider-for-magnitude/archive/refs/heads/main.zip
Readme github.com/devmahbubhussain/ai-provider-for-magnitude ↗

Text generation with tool calling and image input on supported models, running locally on your own computer with Magnitude.

This WordPress plugin is a provider connector for the WordPress AI Client (WordPress 7.0+). It connects your site to a Magnitude server, so any plugin that uses the AI Client can run its prompts on an open-weight model on your own hardware, with no cloud API key.

This is an independent project. It is not affiliated with, endorsed by, or sponsored by Magnitude AI Inc.

How it works

  1. The plugin registers Magnitude with the AI Client. WordPress then lists it on Settings > Connectors by itself.
  2. It reads the models Magnitude is running through Magnitude's OpenAI-compatible API (/inference/v1).
  3. Each model's features follow what Magnitude reports for it. When you download and start a new model, it gets its tools, structured output and image input automatically, with no plugin update.

Requirements

  • WordPress 7.0 or later
  • PHP 7.4 or later
  • The Magnitude app, running with at least one model started. Magnitude lists only models that are running.

Installation

  1. Copy the plugin folder to wp-content/plugins/ai-provider-for-magnitude/, or upload the release zip on Plugins > Add New Plugin > Upload Plugin.
  2. Activate AI Provider for Magnitude.
  3. Open Settings > Magnitude. The status should read Connected, and your running models are listed.

No API key is needed when Magnitude runs on the same computer as WordPress.

Settings

Settings > Magnitude shows the connection status, the server address, and the models Magnitude is running. Check connection asks Magnitude again.

The default address is http://127.0.0.1:10100/inference.

Use Magnitude on another computer

  1. In the Magnitude app on that computer, turn on Network access.
  2. If Require API key is on, copy the key and enter it for Magnitude on Settings > Connectors.
  3. Copy the address under Reachable at, paste it on Settings > Magnitude and save. The plugin adds the /v1 part itself, so either form works.

Only use this on a network you trust.

Fix the address in code

Add this to wp-config.php. The settings field is then locked.

define( 'AI_PROVIDER_FOR_MAGNITUDE_HOST', 'http://192.168.1.20:10100/inference' );

Only http and https addresses with a host and no username or password are accepted. Anything else falls back to the default.

What is supported

Feature Status
Text generation, system instructions, chat history Supported
Tool (function) calling Supported on models that report it
Structured JSON output Supported on models that report it
Image input (vision) Offered on models that report it
Audio input Offered on models that report it
Several answers per request Not available. Magnitude returns one.
Image generation, embeddings, speech Not available. Magnitude does not offer them.

Some models think before they answer. Give those a generous token limit, or they can run out of tokens before the reply.

Hooks

Hook Type Purpose
ai_provider_for_magnitude_host Filter Change the server address. Invalid values fall back to the default.
ai_provider_for_magnitude_request_timeout Filter Change the request timeout in seconds. Default 120.

Development

The plugin has no runtime dependencies. The AI Client ships with WordPress core.

includes/
  Admin/        Plugin list links
  Connection/   Connection status and inspector
  Contracts/    Interfaces
  Host/         Server address resolver
  Http/         Allows the configured local host
  Integration/  Optional WordPress AI plugin integration
  Metadata/     Model list and capability mapping
  Models/       Text generation model
  Provider/     Provider and its registration
  Settings/     Settings screen and address validation
  Traits/       Shared helpers
assets/css/     Settings screen styles
tests/Unit/     PHPUnit tests

Install the dev tools once, then use the composer scripts:

composer install
composer lint      # WordPress Coding Standards, PHP 7.4 compatibility
composer phpstan   # static analysis, level 8
composer test      # PHPUnit unit tests
composer format    # fix coding standard issues automatically

Tooling is configured in phpcs.xml.dist, phpstan.neon.dist and phpunit.xml.dist. The unit tests and PHPStan load the AI Client from the WordPress install that contains the plugin, so keep the plugin in wp-content/plugins. Set AI_MAGNITUDE_ABSPATH to your WordPress root if it lives somewhere else (tests only).

To build a release, copy the plugin without the files listed in .distignore, then run the Plugin Check plugin on the result.

License

GPL-2.0-or-later. See readme.txt for the WordPress.org description and changelog.