🧠 Synapse

AI Agents for All!

Synapse allows you to easily create and manage AI agents in your Laravel application. Inspired by Langchain and Saloon, this package simplifies AI integration and enables scalability.

Installation

Requires PHP 8.1+

Install via Composer:

composer require use-the-fork/synapse

Then, run the installation command:

php artisan synapse:install

If you're not using DatabaseMemory, there's no need to publish the migrations.

Get Started

1. Set up your .env file with the following settings (omit any you don't need):

OPENAI_API_KEY=
OPENAI_API_CHAT_MODEL=gpt-4-turbo
OPENAI_API_EMBEDDING_MODEL=text-embedding-ada-002

ANTHROPIC_API_KEY=
ANTHROPIC_API_CHAT_MODEL=claude-3-5-sonnet-20240620

OLLAMA_BASE_URL=https://foo.bar:1234
OLLAMA_API_CHAT_MODEL=llama3.2
OLLAMA_API_EMBEDDING_MODEL=llama3.2

SERPAPI_API_KEY=
SERPER_API_KEY=
CLEARBIT_API_KEY=
CRUNCHBASE_API_KEY=
FIRECRAWL_API_KEY=

2. Add the service provider if packages aren't autoloaded:

For Laravel 10:

//config/app.php
'providers' => [
    ...
    UseTheFork\Synapse\SynapseServiceProvider::class, 
    ...
];

For Laravel 11:

//bootstrap/providers.php
<?php
return [
    App\Providers\AppServiceProvider::class,
    UseTheFork\Synapse\SynapseServiceProvider::class, 
];

3. If you are not using OpenAI as your default integration:

Open the config/synapse.php file and modify the default integration to the one you prefer:

    'integrations' => [
        'default' => UseTheFork\Synapse\Integrations\OpenAIIntegration::class, 
        'openai' => [
            'key' => env('OPENAI_API_KEY'),
            'chat_model' => env('OPENAI_API_CHAT_MODEL', 'gpt-4-turbo'),
            'embedding_model' => env('OPENAI_API_EMBEDDING_MODEL', 'text-embedding-ada-002'),
        ],
        'claude' => [
            'key' => env('ANTHROPIC_API_KEY'),
            'chat_model' => env('ANTHROPIC_API_CHAT_MODEL', 'claude-3-5-sonnet-20240620'),
        ],
        'ollama' => [
            'base_url' => env('OLLAMA_BASE_URL'),
            'chat_model' => env('OLLAMA_API_CHAT_MODEL', 'llama3.2'),
            'embedding_model' => env('OLLAMA_API_CHAT_MODEL', 'llama3.2'),
        ],
    ],

You're now ready to use Synapse.