itzmekhokan/ai-provider-for-huggingface

AI Provider for Hugging Face for the PHP AI Client SDK. Works as both a Composer package and WordPress plugin.

dev-master 2026-06-20 16:30 UTC

Requires

• php: >=7.4

Requires (Dev)

• dealerdirect/phpcodesniffer-composer-installer: ^1.0
• phpcompatibility/php-compatibility: dev-develop
• phpstan/phpstan: ~2.1
• slevomat/coding-standard: ^8.20
• squizlabs/php_codesniffer: ^3.7 || ^4.0
• wordpress/php-ai-client: ^0.4 || dev-trunk

Suggests

• wordpress/php-ai-client: Required. The core PHP AI Client SDK that this provider extends.

GPL-2.0-or-later

• Khokan Sardar

• wordpress
• router
• ai
• llm
• huggingface

Published on Packagist.org on 2026-07-05 11:49 UTC

README

AI Provider for Hugging Face for the WordPress AI Client. Works as both a Composer package and a WordPress plugin.

This provider registers Hugging Face with the AI Client, exposing the open models served through the Hugging Face Inference Providers router (Llama, Qwen, DeepSeek, Mistral, and more) for text generation through Hugging Face's OpenAI-compatible router API.

Requirements

• PHP 7.4+
• WordPress 6.9+ (when used as a plugin)
• The wordpress/php-ai-client SDK
• A Hugging Face access token from your access tokens page

How it works

The Hugging Face router exposes an OpenAI-compatible API, so this provider is intentionally thin:

Class	Responsibility
Provider\HuggingFaceProvider	Registers the provider, base URL (https://router.huggingface.co/v1), API-key auth, and metadata.
Metadata\HuggingFaceModelMetadataDirectory	Lists models from GET /v1/models and maps capabilities/options.
Models\HuggingFaceTextGenerationModel	Sends POST /v1/chat/completions via the shared OpenAI-compatible base class.

Because the router speaks the Chat Completions format, HuggingFaceTextGenerationModel extends AbstractOpenAiCompatibleTextGenerationModel from the SDK and only overrides request construction.

The Hugging Face router is an aggregator across many inference providers, so the model list reflects whatever chat-capable models the router currently serves. Models are listed alphabetically, with multimodal (image) input enabled for the vision-capable families.

Supported capabilities

• Text generation (chat) with open models routed through Hugging Face
• Chat history / multi-turn conversations
• Multimodal (image) input for vision-capable models
• Function calling and JSON / structured output

Installation

As a WordPress plugin

Copy this directory into wp-content/plugins/ and activate it alongside the WordPress AI Client.

As a Composer package

composer require itzmekhokan/ai-provider-for-huggingface

Author

Khokan Sardar

License

GPL-2.0-or-later