AiWire

AiWire — AI Integration Module for ProcessWire

Connect your ProcessWire site to AI providers: Anthropic (Claude), OpenAI (GPT), Google (Gemini), xAI (Grok), and OpenRouter (400+ models).

Manage multiple API keys per provider, test connections from admin, and use AI in your templates with a clean PHP API.

📖 Full documentation with 25 real-world examples → DOCUMENTATION.md

Features

5 providers — Anthropic, OpenAI, Google, xAI, OpenRouter
Multiple API keys per provider with enable/disable toggle
Automatic fallback — if one key/provider fails, the next one takes over
Connection testing — verify each key with one click in admin
Test Chat — interactive chat with parameter controls (temperature, tokens, model)
File cache — TTL-based caching (day/week/month/year) with page context
Field storage — save AI content to page fields for permanent, editable storage
Content blocks — generate() method for multi-block pages (overview, FAQ, pairings…)
Conversation history — multi-turn chat support
Cost optimization — route tasks to different providers/models by complexity
Logging — standard + debug logging via ProcessWire log system

Requirements

PHP 8.1+
ProcessWire 3.0.210+
cURL extension enabled

Installation

Download or clone into site/modules/AiWire/
Admin → Modules → Refresh → Install AiWire
Configure → add API keys → click test button
Use $modules->get('AiWire') in your templates

site/modules/AiWire/
├── AiWire.module.php       # Main module
├── AiWireProvider.php       # API client for all providers
├── AiWireCache.php          # File-based cache system
├── models.json              # Editable provider model list
├── README.md                # This file
├── DOCUMENTATION.md         # Full API reference + 25 examples
└── LICENSE

Updating Models

Provider model IDs and labels are loaded from models.json. Edit that file when a provider adds, renames, or retires a model:

{
  "openai": {
    "defaultModel": "gpt-5.4",
    "models": {
      "gpt-5.4": "GPT-5.4",
      "gpt-5.4-mini": "GPT-5.4 Mini"
    }
  }
}

If models.json is missing or invalid, AiWire falls back to the built-in model list.

OpenAI and OpenRouter also support model refresh from the module settings. Refreshed models are stored in module config separately from models.json and are used first; if no refreshed list exists, AiWire falls back to models.json, then to built-in defaults.

Each API key can also define a custom model. The custom model value is optional, but when present it overrides the selected model for that key. This is useful for private, preview, account-gated, proxy, or newly released model IDs.

API Key Storage

Keys can be stored directly in the module settings, or as environment references such as env:OPENAI_API_KEY. Environment references avoid saving the real secret in ProcessWire module config.

Getting API Keys

Provider	Where to get a key
Anthropic	console.anthropic.com
OpenAI	platform.openai.com/api-keys
Google	aistudio.google.com/apikey
xAI	console.x.ai
OpenRouter	openrouter.ai/keys

Quick Start

$ai = $modules->get('AiWire');

// Simple — returns text only
echo $ai->chat('What is ProcessWire CMS?');

// Full response with metadata
$result = $ai->ask('Explain quantum computing in simple terms');
if ($result['success']) {
    echo $result['content'];              // AI response text
    echo $result['usage']['total_tokens']; // tokens used
}

// Fallback — tries Anthropic → OpenAI → Google
$result = $ai->askWithFallback('Summarize this article...', [
    'provider'          => 'anthropic',
    'fallbackProviders' => ['openai', 'google'],
]);

// Generate multiple AI blocks for a product page
$results = $ai->generate($page, [
    ['field' => 'ai_overview',    'prompt' => "Write overview of {$page->title}..."],
    ['field' => 'ai_food_pairing','prompt' => "Suggest pairings for {$page->title}..."],
    ['field' => 'ai_brand_story', 'prompt' => "Share 3 facts about {$page->brand->title}..."],
], ['cache' => 'M', 'temperature' => 0.7]);

API Methods

Method	Returns	Description
`chat($msg, $opts)`	`string`	Simple text response, empty on error
`ask($msg, $opts)`	`array`	Full response with `success`, `content`, `usage`, `raw`
`askWithFallback($msg, $opts)`	`array`	Tries all keys/providers until success
`askMultiple($msg, $providers)`	`array`	Same prompt to multiple providers
`askAndSave($page, $fields, $msg)`	`array`	Ask AI + save to page field (skip if exists)
`generate($page, $blocks, $opts)`	`array`	Multi-block generation with per-block settings
`saveTo($page, $field, $content)`	`bool`	Save content to page field
`loadFrom($page, $field)`	`?string`	Load content from page field
`getProvider($key, $specific, $idx)`	`?Provider`	Get raw provider instance
`getProvidersStatus()`	`array`	Status of all providers and keys
`clearCache($page)`	`int`	Clear cache for a page
`clearAllCache()`	`int`	Clear entire cache
`cacheStats()`	`array`	Cache files count, size, etc.

→ Full API reference: DOCUMENTATION.md

Result Format

// Successful response from ask(), askWithFallback(), askAndSave(), generate()
[
    'success' => true,
    'content' => 'The AI response text...',
    'message' => 'OK',
    'usage'   => ['input_tokens' => 25, 'output_tokens' => 148, 'total_tokens' => 173],
    'raw'     => [ /* full API response */ ],
    'cached'  => false,       // true if served from file cache
    'source'  => 'ai',        // askAndSave/generate only: 'ai', 'field', or 'error'
]

Options

Option	Type	Default	Description
`provider`	string	Module default	`anthropic`, `openai`, `google`, `xai`, `openrouter`
`model`	string	Key’s model	Override model for this call
`systemPrompt`	string	Module default	System instructions for the AI
`maxTokens`	int	1024	Max tokens in response
`temperature`	float	0.7	0.0 = precise, 1.0+ = creative
`history`	array	`[]`	Previous messages for multi-turn chat
`keyIndex`	int	—	Use a specific key by index (0-based)
`fallbackProviders`	array	—	Fallback providers for `askWithFallback`
`cache`	string\|bool	`false`	TTL: `'D'`, `'W'`, `'M'`, `'Y'`, `'2W'`, `'3M'`
`pageId`	int\|Page	0	Page context for cache scoping
`timeout`	int	30	Request timeout in seconds
`overwrite`	bool	false	`askAndSave`: always call AI even if field has content
`quiet`	bool	true	`askAndSave`: save without triggering PW hooks

Supported Models

The default editable list in models.json currently includes:

Provider	Models
Anthropic	`claude-opus-4-7`, `claude-opus-4-6`, `claude-sonnet-4-6-20260217`, `claude-haiku-4-5-20251001`
OpenAI	`gpt-5.4`, `gpt-5.4-mini`, `gpt-5.4-nano`, `gpt-5.2`, `gpt-4.1`
Google	`gemini-3.1-pro-preview`, `gemini-3-flash`, `gemini-3.1-flash-lite`, `gemini-2.5-flash`
xAI	`grok-4.20`, `grok-4-1-fast-reasoning`, `grok-4-1-fast-non-reasoning`, `grok-code-fast-1`
OpenRouter	`deepseek/deepseek-v3.2`, `qwen/qwen3-max-thinking`, `meta-llama/llama-4-maverick`, `amazon/nova-2-lite-v1`, `z-ai/glm-5`, `xiaomi/mimo-v2-flash`, `bytedance-seed/seed-1.6`, and 400+ more

Documentation & Examples

Full documentation with detailed API reference, 25 real-world examples (with ProcessWire setup, code, and expected output), cache strategies, field storage, multi-provider pipelines, and best practices:

→ DOCUMENTATION.md

Content generation — product pages, brand enrichment, category descriptions, cocktail recipes, region guides, tasting notes, review summaries

SEO & translations — auto-generate meta descriptions, OG titles, multi-language content

Chatbots & APIs — AI sommelier, gift recommendations, product comparison, form routing

Infrastructure — multi-provider cost optimization, fallback chains, key rotation, cache strategies, provider monitoring, team key separation, bulk LazyCron generation

License

MIT — free for personal and commercial use.

Author

Maxim Alex — smnv.org — maxim@smnv.org

Built for the ProcessWire community.

This site is open source. Improve this page.