Symfony AI

https://github.com/symfony/ai
Admin
Symfony AI provides a set of components and bundles to integrate AI capabilities into PHP...
Tokens:283,271
Snippets:2,680
Trust Score:9.5
License:MIT
Update:3 weeks ago
Show doc for...
Tokens:
Raw
### Setup commands

Source: https://github.com/symfony/ai/blob/main/examples/AGENTS.md

Install dependencies, link local AI components, and start Docker for store examples.

```bash
composer install
../link  # Link local AI components
docker compose up -d  # For store examples
```

--------------------------------

### Start Docker services for store examples

Source: https://github.com/symfony/ai/blob/main/examples/README.md

Start the required Docker services for store examples that need locally running services.

```bash
docker compose up -d
```

--------------------------------

### Run example with verbose output

Source: https://github.com/symfony/ai/blob/main/examples/README.md

Run an example with -vv or -vvv to get more insights into HTTP and tool calls.

```bash
php openai/toolcall-stream.php -vvv
```

--------------------------------

### Start the demo application

Source: https://github.com/symfony/ai/blob/main/AGENTS.md

Install dependencies and start the demo Symfony application. The demo is a full Symfony application.

```bash
cd demo
composer install
symfony server:start
```

--------------------------------

### Setup commands for Symfony AI demo

Source: https://github.com/symfony/ai/blob/main/demo/AGENTS.md

Setup commands for the Symfony AI demo: start PostgreSQL with pgvector, install dependencies, set the OpenAI API key (HUGGINGFACE_API_KEY only if wiring HF agents), set up the vector store before indexing, index and retrieve from the blog store, and serve the app at https://localhost:8000/.

```bash
docker compose up -d                          # Start PostgreSQL (pgvector)
composer install
echo "OPENAI_API_KEY='sk-...'" > .env.local   # HUGGINGFACE_API_KEY only if you wire HF agents

# Vector store must be set up before indexing
symfony console ai:store:setup ai.store.postgres.symfony_blog
symfony console ai:store:index blog -vv
symfony console ai:store:retrieve blog "Week of Symfony"

symfony serve -d                              # https://localhost:8000/
```

--------------------------------

### Set up Symfony app: clone, install, start Docker, serve

Source: https://github.com/symfony/ai/blob/main/demo/README.md

Clones the repository, installs dependencies, starts the Docker environment, and serves the Symfony app. After running, open https://localhost:8000/ in your browser; you may need to bypass a self-signed certificate warning.

```shell
git clone git@github.com:symfony/ai-demo.git
cd ai-demo
composer install
docker compose up -d
symfony serve -d
```

--------------------------------

### ai:store:setup command usage

Source: https://github.com/symfony/ai/blob/main/docs/bundles/ai-bundle.rst

Terminal commands to set up a store, with an example for the chromadb store.

```terminal
$ php bin/console ai:store:setup <store>

# Setup the chromadb store
$ php bin/console ai:store:setup ai.store.chromadb.default
```

--------------------------------

### Change to examples directory

Source: https://github.com/symfony/ai/blob/main/examples/README.md

Navigate into the examples directory before running any commands.

```bash
cd examples
```

--------------------------------

### Run all examples or a specific example

Source: https://github.com/symfony/ai/blob/main/examples/huggingface/README.md

Commands to run all examples via the runner or a specific example standalone, e.g., object detection.

```bash
# Run all example with runner:
./runner huggingface

# Or run a specific example standalone, e.g., object detection:
php huggingface/object-detection.php
```

--------------------------------

### Install and configure OpenSearch vector store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the OpenSearch store bridge, then create a store with StoreFactory::create() and configure it in config/packages/ai.yaml. The index is created by setup(). The endpoint can be replaced by a pre-scoped HTTP client service.

```terminal
$ composer require symfony/ai-open-search-store
```

```php
use Symfony\AI\Store\Bridge\OpenSearch\StoreFactory;

$store = StoreFactory::create(
    'my_documents',
    'https://localhost:9200',
    vectorsField: '_vectors',
    dimensions: 1536,
    spaceType: 'cosinesimil',
);
```

```yaml
# config/packages/ai.yaml
ai:
    store:
        opensearch:
            my_store:
                endpoint: '%env(OPENSEARCH_URL)%'
                # or a pre-scoped HTTP client service instead of an endpoint:
                # http_client: 'app.opensearch_client'
                index_name: 'my_documents'
                vectors_field: '_vectors'
                dimensions: 1536
                space_type: 'cosinesimil'
```

--------------------------------

### Install and configure Typesense vector store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the Typesense store bridge, then create a store with StoreFactory::create() and configure it in config/packages/ai.yaml. The collection is created by setup().

```terminal
$ composer require symfony/ai-typesense-store
```

```php
use Symfony\AI\Store\Bridge\Typesense\StoreFactory;

$store = StoreFactory::create(
    'documents',
    'http://localhost:8108',
    'your-api-key',
    vectorFieldName: '_vectors',
    embeddingsDimension: 1536,
);
```

```yaml
# config/packages/ai.yaml
ai:
    store:
        typesense:
            my_store:
                endpoint: '%env(TYPESENSE_URL)%'
                api_key: '%env(TYPESENSE_API_KEY)%'
                collection: 'documents'
                vector_field: '_vectors'
                dimensions: 1536
```

--------------------------------

### Install models.dev bridge and data package

Source: https://github.com/symfony/ai/blob/main/docs/components/platform/model-catalogs.rst

Installs the models.dev bridge and the daily snapshot data package. Run this to get the continuously updated catalog.

```bash
composer require symfony/ai-models-dev-platform symfony/models-dev
```

--------------------------------

### Get Model Information Examples

Source: https://github.com/symfony/ai/blob/main/src/platform/src/Bridge/HuggingFace/README.md

Examples of retrieving information for specific models, such as meta-llama/Llama-2-7b-chat and gpt2.

```bash
# Get information about a specific model
ai:huggingface:model-info meta-llama/Llama-2-7b-chat
```

```bash
ai:huggingface:model-info gpt2
```

--------------------------------

### Run MiniMax text-to-speech examples

Source: https://github.com/symfony/ai/blob/main/examples/minimax/README.md

Run the MiniMax text-to-speech examples. The synchronous example outputs audio as binary, so pipe it to a player like mpg123. The asynchronous example saves audio to minimax/minimax-speech.mp3.

```bash
php minimax/text-to-speech.php | mpg123 -
php minimax/text-to-speech-async.php
```

--------------------------------

### Run MiniMax image, music, and video examples

Source: https://github.com/symfony/ai/blob/main/examples/minimax/README.md

Run the MiniMax image, music, and video generation examples. The music example outputs audio as binary, so pipe it to a player like mpg123.

```bash
php minimax/text-to-image.php
php minimax/music.php | mpg123 -
php minimax/text-to-video.php
```

--------------------------------

### Install community extension

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/extensions.rst

Install a community extension, using the PHPUnit extension as an example.

```terminal
$ composer require --dev matesofmate/phpunit-extension
```

--------------------------------

### Install Symfony AI Bundle and Agent

Source: https://github.com/symfony/ai/blob/main/docs/index.rst

Install the AI Bundle and the Agent component to get started with Symfony AI.

```terminal
$ composer require symfony/ai-bundle symfony/ai-agent
```

--------------------------------

### Run ElevenLabs text-to-speech example with mpg123

Source: https://github.com/symfony/ai/blob/main/examples/elevenlabs/README.md

Runs the text-to-speech example and pipes its output to mpg123 for audio playback. Requires the mpg123 tool to be installed.

```bash
php elevenlabs/text-to-speech.php | mpg123 -
```

--------------------------------

### Run a standalone example

Source: https://github.com/symfony/ai/blob/main/examples/README.md

Execute a standalone PHP example script from the command line.

```bash
php openai/chat.php
```

--------------------------------

### Run the example runner

Source: https://github.com/symfony/ai/blob/main/examples/README.md

Execute the example runner to run examples in parallel sub-processes.

```bash
./runner
```

--------------------------------

### Install and configure Meilisearch vector store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the Meilisearch store bridge, then create a store with StoreFactory::create() and configure it in config/packages/ai.yaml. The index and its embedder settings are created by setup().

```terminal
$ composer require symfony/ai-meilisearch-store
```

```php
use Symfony\AI\Store\Bridge\Meilisearch\StoreFactory;

$store = StoreFactory::create(
    'documents',
    'http://localhost:7700',
    'your-api-key',
    embedder: 'default',
    vectorFieldName: '_vectors',
    embeddingsDimension: 1536,
    semanticRatio: 1.0,
);
```

```yaml
# config/packages/ai.yaml
ai:
    store:
        meilisearch:
            my_store:
                endpoint: '%env(MEILISEARCH_URL)%'
                api_key: '%env(MEILISEARCH_API_KEY)%'
                index_name: 'documents'
                embedder: 'default'
                vector_field: '_vectors'
                dimensions: 1536
                semantic_ratio: 1.0
```

--------------------------------

### Download models and start Ollama server

Source: https://github.com/symfony/ai/blob/main/examples/ollama/README.md

First, download the required models (llama3.2 and nomic-embed-text) using the ollama pull commands. Then start the Ollama server with ollama serve. The examples expect Ollama to run on localhost:11434 and use these models.

```bash
ollama pull llama3.2
ollama pull nomic-embed-text
```

```bash
ollama serve
```

--------------------------------

### Run self-contained examples

Source: https://github.com/symfony/ai/blob/main/AGENTS.md

Run the self-contained examples from the examples directory. Many examples require environment variables (see .env files in example directories).

```bash
cd examples
php anthropic/chat.php
php openai/toolcall.php
```

--------------------------------

### Install and configure Elasticsearch vector store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the Elasticsearch store bridge, then create a store with StoreFactory::create() and configure it in config/packages/ai.yaml. The index is created by setup(). The endpoint can be replaced by a pre-scoped HTTP client service.

```terminal
$ composer require symfony/ai-elasticsearch-store
```

```php
use Symfony\AI\Store\Bridge\Elasticsearch\StoreFactory;

$store = StoreFactory::create(
    'my_documents',
    'https://localhost:9200',
    vectorsField: '_vectors',
    dimensions: 1536,
    similarity: 'cosine',
);
```

```yaml
# config/packages/ai.yaml
ai:
    store:
        elasticsearch:
            my_store:
                endpoint: '%env(ELASTICSEARCH_URL)%'
                # or a pre-scoped HTTP client service instead of an endpoint:
                # http_client: 'app.elasticsearch_client'
                index_name: 'my_documents'
                vectors_field: '_vectors'
                dimensions: 1536
                similarity: 'cosine'
```

--------------------------------

### Install and initialize Mate

Source: https://github.com/symfony/ai/blob/main/src/mate/README.md

Install the Mate component as a dev dependency, initialize it in the project, and refresh the autoloader. Run these commands in order to set up the command-line assistant.

```bash
composer require --dev symfony/ai-mate
vendor/bin/mate init
composer dump-autoload
```

--------------------------------

### Setup Vector Store with $store->setup()

Source: https://github.com/symfony/ai/blob/main/docs/components/store/s3-vectors.rst

Initializes the vector bucket and index before use. The dimension is required; distanceMetric, dataType, encryption, and tags are optional.

```php
// Setup the vector bucket and index
$store->setup([
    'dimension' => 1536,
    'distanceMetric' => \AsyncAws\S3Vectors\Enum\DistanceMetric::COSINE, // Optional
    'dataType' => \AsyncAws\S3Vectors\Enum\DataType::FLOAT_32, // Optional
    'encryption' => ['kmsKeyId' => 'your-kms-key-id'], // Optional
    'tags' => ['env' => 'production'], // Optional
]);
```

--------------------------------

### Install Mate via Composer

Source: https://github.com/symfony/ai/blob/main/src/mate/README.md

Install the Mate component as a dev dependency. This is the installation command shown in the Installation section.

```bash
composer require --dev symfony/ai-mate
```

--------------------------------

### Implement ManagedStoreInterface with setup() and drop()

Source: https://github.com/symfony/ai/blob/main/docs/components/chat.rst

Implements ManagedStoreInterface along with MessageStoreInterface to support setup and drop operations. The setup() method creates the store, and drop() removes the store and its messages.

```php
use Symfony\AI\Chat\ManagedStoreInterface;
use Symfony\AI\Chat\MessageStoreInterface;

class MyCustomStore implements ManagedStoreInterface, MessageStoreInterface
{
    # ...

    public function setup(array $options = []): void
    {
        // Implementation to create the store
    }

    public function drop(): void
    {
        // Implementation to drop the store (and related messages)
    }
}
```

--------------------------------

### Running examples

Source: https://github.com/symfony/ai/blob/main/examples/AGENTS.md

Run individual examples or batch execution with the runner script. Use filters to select specific platforms or patterns.

```bash
# Single example
php openai/chat.php
php openai/toolcall-stream.php -vvv

# Batch execution
./runner  # All examples
./runner openai mistral  # Specific platforms
./runner --filter=toolcall  # Pattern filter
```

--------------------------------

### Run MiniMax chat examples

Source: https://github.com/symfony/ai/blob/main/examples/minimax/README.md

Run the MiniMax chat examples: standard chat, streaming chat, and chat with token usage. Requires MINI_MAX_API_KEY in examples/.env.local.

```bash
php minimax/chat.php
php minimax/chat-as-stream.php
php minimax/chat-with-token-usage.php
```

--------------------------------

### Install and verify a Mate extension

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/creating-extensions.rst

Install the extension and verify its capabilities. In a project that ran mate init, the Composer plugin runs mate discover after install, adding the extension to mate/extensions.php enabled by default. Without the plugin, run vendor/bin/mate discover yourself.

```terminal
$ composer require --dev vendor/my-extension
$ vendor/bin/mate debug:capabilities --extension=vendor/my-extension
```

--------------------------------

### Install and configure ManticoreSearch vector store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the ManticoreSearch store bridge, then create a store with StoreFactory::create() and configure it in config/packages/ai.yaml. The table is created by setup(). The endpoint can be replaced by a pre-scoped HTTP client service.

```terminal
$ composer require symfony/ai-manticore-search-store
```

```php
use Symfony\AI\Store\Bridge\ManticoreSearch\StoreFactory;

$store = StoreFactory::create(
    'documents',
    'http://localhost:9308',
    field: '_vectors',
    type: 'hnsw',
    similarity: 'cosine',
    dimensions: 1536,
    quantization: '8bit',
);
```

```yaml
# config/packages/ai.yaml
ai:
    store:
        manticoresearch:
            my_store:
                endpoint: '%env(MANTICORESEARCH_URL)%'
                # or a pre-scoped HTTP client service instead of an endpoint:
                # http_client: 'app.manticoresearch_client'
                table: 'documents'
                field: '_vectors'
                type: 'hnsw'
                similarity: 'cosine'
                dimensions: 1536
                quantization: '8bit'
```

--------------------------------

### Install and initialize AI Mate

Source: https://github.com/symfony/ai/blob/main/demo/README.md

Install AI Mate as a dev dependency, initialize its configuration, and discover available tools. Run these commands in the project root.

```shell
# Install AI Mate
composer require --dev symfony/ai-mate

# Initialize configuration
vendor/bin/mate init

# Discover available tools
vendor/bin/mate discover
```

--------------------------------

### Replay recorded examples

Source: https://github.com/symfony/ai/blob/main/examples/README.md

Replay all recorded examples offline and compare against goldens; no API keys needed.

```bash
vendor/bin/phpunit
```

--------------------------------

### Implement ManagedStoreInterface and StoreInterface

Source: https://github.com/symfony/ai/blob/main/docs/components/store.rst

Shows how to implement both ManagedStoreInterface and StoreInterface to support setup and drop operations. The setup method creates the store, and the drop method removes it along with related vectors.

```php
use Symfony\AI\Store\ManagedStoreInterface;
use Symfony\AI\Store\StoreInterface;

class MyCustomStore implements ManagedStoreInterface, StoreInterface
{
    # ...

    public function setup(array $options = []): void
    {
        // Implementation to create the store
    }

    public function drop(array $options = []): void
    {
        // Implementation to drop the store (and related vectors)
    }
}
```

--------------------------------

### Run Venice examples scripts

Source: https://github.com/symfony/ai/blob/main/examples/venice/README.md

Shows the commands to execute the provided Venice example scripts: chat, stream, text-to-image, and text-to-video. All require VENICE_API_KEY.

```bash
php venice/chat.php
php venice/stream.php
php venice/text-to-image.php
php venice/text-to-video.php
```

--------------------------------

### Install Mate extensions with Composer

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/extensions.rst

Install the two Mate extensions as dev dependencies. The Composer plugin then runs 'mate discover' to enable the extensions and install their skills.

```terminal
$ composer require --dev symfony/ai-symfony-mate-extension symfony/ai-monolog-mate-extension
```

--------------------------------

### Install built-in tools

Source: https://github.com/symfony/ai/blob/main/ai.symfony.com/templates/cookbook/content/ai-bundle-fast-track.html

Use Composer to install standalone tool packages.

```bash
$ composer require symfony/ai-brave-tool
$ composer require symfony/ai-wikipedia-tool
$ composer require symfony/ai-open-meteo-tool
```

--------------------------------

### Install Clock Tool

Source: https://github.com/symfony/ai/blob/main/docs/components/agent/tools.rst

Install the Clock tool. No API key is required.

```terminal
$ composer require symfony/ai-clock-tool
```

--------------------------------

### Install and Use MCP Tool

Source: https://github.com/symfony/ai/blob/main/docs/components/agent/tools.rst

Install the MCP tool and then create a toolset from a remote MCP server, wrapping it in an McpToolbox for the agent. The server name prefixes each tool, e.g. filesystem_read_file. A server that cannot be reached contributes no tools and is retried after retryAfter seconds (default 60).

```terminal
$ composer require symfony/ai-mcp-tool
```

```php
use Mcp\Client;
use Mcp\Client\Transport\StdioTransport;
use Symfony\AI\Agent\Agent;
use Symfony\AI\Agent\Bridge\Mcp\ClientToolset;
use Symfony\AI\Agent\Bridge\Mcp\McpToolbox;

$toolset = new ClientToolset(
    'filesystem',
    Client::builder()->build(),
    new StdioTransport('npx', ['-y', '@modelcontextprotocol/server-filesystem', '/tmp']),
);

$agent = new Agent($platform, 'gpt-4o-mini', toolbox: new McpToolbox($toolset));
```

--------------------------------

### Install with Composer: symfony/ai-platform

Source: https://github.com/symfony/ai/blob/main/src/platform/README.md

Use this command to install the core Platform component. Requires Composer and a PHP environment.

```bash
composer require symfony/ai-platform
```

--------------------------------

### Install symfony/ai-neo4j-store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the Neo4j vector store bridge package.

```terminal
$ composer require symfony/ai-neo4j-store
```

--------------------------------

### Install symfony/ai-surreal-db-store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the SurrealDB vector store bridge package.

```terminal
$ composer require symfony/ai-surreal-db-store
```

--------------------------------

### Install root dependencies

Source: https://github.com/symfony/ai/blob/main/CONTRIBUTING.md

Run this command in the repository root to install dependencies, enabling the use of the ./link script.

```bash
composer install
```

--------------------------------

### Install symfony/ai-vertex-ai-platform

Source: https://github.com/symfony/ai/blob/main/docs/components/platform/vertexai.rst

Install the Symfony AI Vertex AI platform component using Composer.

```terminal
$ composer require symfony/ai-vertex-ai-platform
```

--------------------------------

### Install SQLite store with Composer

Source: https://github.com/symfony/ai/blob/main/docs/components/store/sqlite.rst

Install the symfony/ai-sqlite-store package using Composer.

```terminal
$ composer require symfony/ai-sqlite-store
```

--------------------------------

### Install Filesystem Tool

Source: https://github.com/symfony/ai/blob/main/docs/components/agent/tools.rst

Install the Filesystem tool. Restrict the basePath to an allowed directory and note that allowDelete is disabled by default.

```terminal
$ composer require symfony/ai-filesystem-tool
```

--------------------------------

### List Models Examples

Source: https://github.com/symfony/ai/blob/main/src/platform/src/Bridge/HuggingFace/README.md

Examples of using the model-list command with different filters: by task, by provider with warm inference, and by search term.

```bash
# List all text generation models
ai:huggingface:model-list --task=text-generation
```

```bash
# List warm models for a specific provider
ai:huggingface:model-list --provider=hf-inference --warm
```

```bash
# Search for specific models
ai:huggingface:model-list --search=llama
```

--------------------------------

### Filter runner examples

Source: https://github.com/symfony/ai/blob/main/examples/README.md

Use the --filter option to run a specific subset of examples.

```bash
./runner --filter=toolcall
```

--------------------------------

### Install Symfony AI components

Source: https://github.com/symfony/ai/blob/main/ai.symfony.com/templates/cookbook/content/human-in-the-loop.html

Install the required Symfony AI Platform, Agent, and EventDispatcher components.

```bash
composer require symfony/ai-platform symfony/ai-agent symfony/event-dispatcher
```

--------------------------------

### Install SimilaritySearch Tool

Source: https://github.com/symfony/ai/blob/main/docs/components/agent/tools.rst

Install the SimilaritySearch tool for RAG. Requires a vectorizer and a configured vector store.

```terminal
$ composer require symfony/ai-similarity-search-tool
```

--------------------------------

### Start MCP server with Symfony console

Source: https://github.com/symfony/ai/blob/main/demo/README.md

Starts the MCP server using the Symfony console command. Use this to test the server locally.

```shell
symfony console mcp:server
```

--------------------------------

### Install YouTube Transcriber Tool

Source: https://github.com/symfony/ai/blob/main/docs/components/agent/tools.rst

Install the YouTube Transcriber tool. No API key is required.

```terminal
$ composer require symfony/ai-youtube-tool
```

--------------------------------

### Run Higgsfield text-to-image and image-to-video examples

Source: https://github.com/symfony/ai/blob/main/examples/higgsfield/README.md

Commands to execute the example scripts. Requires HIGGSFIELD_API_KEY and HIGGSFIELD_API_SECRET in .env.local, provided as a single <key-id>:<key-secret> pair split at the colon. Uses short aliases like soul-2 and kling-2.5-i2v from the default CuratedModelCatalog.

```bash
php higgsfield/text-to-image.php
php higgsfield/image-to-video.php
```

--------------------------------

### Example text file for system prompt

Source: https://github.com/symfony/ai/blob/main/docs/bundles/ai-bundle.rst

Example of a text file used as a system prompt. The file can be in any text format (.txt, .json, .md, etc.). The content is used verbatim as the system prompt.

```text
You are a helpful and knowledgeable assistant.

Guidelines:
- Be clear and direct in your responses
- Provide examples when appropriate
- Be respectful and professional at all times
```

--------------------------------

### Set up store and index documents via console

Source: https://github.com/symfony/ai/blob/main/ai.symfony.com/templates/cookbook/content/ai-bundle-fast-track.html

Console commands to prepare the ChromaDB store and index a document. The first command sets up the store infrastructure, and the second indexes a document from a specified path.

```bash
$ php bin/console ai:store:setup ai.store.chromadb.knowledge_base
$ php bin/console ai:store:index docs --source=/path/to/document.txt
```

--------------------------------

### Install symfony/ai-maria-db-store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the MariaDB vector store bridge. Requires MariaDB 11.7+ and PHP ext-pdo.

```terminal
$ composer require symfony/ai-maria-db-store
```

--------------------------------

### Install symfony/ai-supabase-store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the Supabase vector store bridge. Unlike the Postgres store, Supabase requires manual setup of the database schema because it does not allow arbitrary SQL execution through the REST API.

```terminal
$ composer require symfony/ai-supabase-store
```

--------------------------------

### Add a tool dynamically (partial example)

Source: https://github.com/symfony/ai/blob/main/docs/cookbook/dynamic-tools.rst

Starts the example for adding a new 'echo' tool dynamically. The source shows the imports and the beginning of the getTools() method, but the full implementation is not displayed on this page.

```php
use Symfony\AI\Platform\Tool\ExecutionReference;
use Symfony\AI\Platform\Tool\Tool;

// ...existing code...
public function getTools(): array
{
    $tools = $this->innerToolbox->getTools();

    // ... (rest of the method is not shown in the source)
}
```

--------------------------------

### Install Mate with Composer

Source: https://github.com/symfony/ai/blob/main/docs/components/mate.rst

Install the core Mate package as a development dependency using Composer. This is the first step to use Mate in any PHP project.

```terminal
$ composer require --dev symfony/ai-mate
```

--------------------------------

### Register namespace and discover extensions

Source: https://github.com/symfony/ai/blob/main/docs/components/mate.rst

After configuring composer.json, run these commands to regenerate the autoloader and let Mate discover installed extensions. 'mate discover' registers extensions, writes agent instructions, and installs skills.

```terminal
$ composer dump-autoload
$ vendor/bin/mate discover
```

--------------------------------

### ai:agent:call command usage

Source: https://github.com/symfony/ai/blob/main/docs/bundles/ai-bundle.rst

Terminal commands to start an interactive chat with an agent, including examples for default and specific agents.

```terminal
$ php bin/console ai:agent:call <agent>

# Chat with the default agent
$ php bin/console ai:agent:call default

# Chat with a specific agent
$ php bin/console ai:agent:call wikipedia
```

--------------------------------

### Initialize Mate in your project

Source: https://github.com/symfony/ai/blob/main/docs/components/mate.rst

Run 'mate init' to start using Mate. It asks which command your coding agent should use to run Mate, and creates the necessary configuration files and directories.

```terminal
$ vendor/bin/mate init
```

--------------------------------

### Run multiple executions concurrently

Source: https://github.com/symfony/ai/blob/main/docs/components/agent.rst

Start multiple executions and advance them together to have their requests in flight simultaneously. This example shows how to iterate two executions side by side.

```php
$first = $agent->call('Summarize this page.')->getIterator();
$second = $agent->call('Translate this page.')->getIterator();

// running each generator to its first update sends both requests
$first->current();
$second->current();

while ($first->valid() || $second->valid()) {
    $first->next();
    $second->next();
}
```

--------------------------------

### Example JSON file for system prompt

Source: https://github.com/symfony/ai/blob/main/docs/bundles/ai-bundle.rst

Example of a JSON file used as a system prompt. The file can be in any text format, including JSON. The content is used as the system prompt.

```json
{
  "role": "You are an expert code reviewer",
  "responsibilities": [
    "Review code for bugs and potential issues",
    "Suggest improvements for code quality"
  ],
  "tone": "constructive and educational"
}
```

--------------------------------

### Register custom tools with AsTool attribute

Source: https://github.com/symfony/ai/blob/main/docs/bundles/ai-bundle.rst

Shows the start of registering a custom tool using the AsTool attribute. The example includes the use statement for the attribute class.

```php
use Symfony\AI\Agent\Toolbox\Attribute\AsTool;
```

--------------------------------

### Postgres store standalone setup with PDO and DBAL

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Standalone creation of a Postgres store using PDO or a Doctrine DBAL connection. The table and index are created by setup(). Available distances: Distance::Cosine, Distance::InnerProduct, Distance::L1, Distance::L2 (default).

```php
use Symfony\AI\Store\Bridge\Postgres\Distance;
use Symfony\AI\Store\Bridge\Postgres\StoreFactory;

$pdo = new \PDO('pgsql:host=localhost;dbname=mydb', $user, $password);
$store = StoreFactory::createStoreFromPdo($pdo, 'documents', vectorFieldName: 'embedding', distance: Distance::Cosine);

// or from a Doctrine DBAL connection
$store = StoreFactory::createStoreFromDbal($connection, 'documents', vectorFieldName: 'embedding', distance: Distance::Cosine);
```

--------------------------------

### Install Symfony AI Platform and Agent via Composer

Source: https://github.com/symfony/ai/blob/main/ai.symfony.com/templates/cookbook/content/chatbot-with-memory.html

Run this Composer command to install the required Symfony AI Platform and Agent components.

```bash
composer require symfony/ai-platform symfony/ai-agent
```

--------------------------------

### Using MemoryInputProcessor with StaticMemoryProvider

Source: https://github.com/symfony/ai/blob/main/docs/components/agent.rst

Demonstrates setting up memory with a StaticMemoryProvider and MemoryInputProcessor, then creating an agent with the memory processor and calling it.

```php
use Symfony\AI\Agent\Agent;
use Symfony\AI\Agent\Memory\MemoryInputProcessor;
use Symfony\AI\Agent\Memory\StaticMemoryProvider;
use Symfony\AI\Platform\Message\Message;
use Symfony\AI\Platform\Message\MessageBag;

// Platform instantiation

$personalFacts = new StaticMemoryProvider([
    'My name is Wilhelm Tell',
    'I wish to be a swiss national hero',
    'I am struggling with hitting apples but want to be professional with the bow and arrow',
]);
$memoryProcessor = new MemoryInputProcessor([$personalFacts]);

$agent = new Agent($platform, $model, [$memoryProcessor]);
$messages = new MessageBag(Message::ofUser('What do we do today?'));
$result = $agent->call($messages);
```

--------------------------------

### Initialize In-Memory Vector Store

Source: https://github.com/symfony/ai/blob/main/ai.symfony.com/templates/cookbook/content/rag-implementation.html

Creates an in-memory vector store for holding embeddings. Ideal for getting started; see the 'Going to Production' section for persistent stores.

```php
use Symfony\AI\Store\InMemory\Store;

$store = new Store();
```

--------------------------------

### Run Decart examples

Source: https://github.com/symfony/ai/blob/main/examples/decart/README.md

Commands to run the Decart example scripts for text-to-image, text-to-video, animated image, and image editing. Each script saves its result next to the script.

```bash
php decart/text-to-image.php
php decart/text-to-video.php
php decart/animated-image.php
php decart/image-editing.php
```

--------------------------------

### Development Commands

Source: https://github.com/symfony/ai/blob/main/ai.symfony.com/AGENTS.md

Development commands for the ai.symfony.com website: install dependencies, start the server, clear cache, list asset paths, and add JS/CSS dependencies via importmap.

```bash
# Install dependencies
composer install

# Start development server
symfony server:start

# Clear cache
bin/console cache:clear

# List available asset paths
bin/console debug:asset-map

# Add a new JS/CSS dependency via importmap
bin/console importmap:require <package>
```

--------------------------------

### Instantiate Agent with processors

Source: https://github.com/symfony/ai/blob/main/docs/components/agent.rst

Shows how to create an Agent instance with platform, model, and input/output processors. The processors are provided as arguments.

```php
use Symfony\AI\Agent\Agent;

// Initialize Platform, LLM and processors

$agent = new Agent($platform, $model, $inputProcessors, $outputProcessors);
```

--------------------------------

### ReadAllowPolicy auto-allows read operations

Source: https://github.com/symfony/ai/blob/main/ai.symfony.com/templates/cookbook/content/human-in-the-loop.html

Example policy that auto-allows tool calls whose names contain read-like patterns (read, get, list, search, find, show) and asks the user otherwise.

```php
use Symfony\AI\Platform\Result\ToolCall;

class ReadAllowPolicy implements PolicyInterface
{
    public function decide(ToolCall $toolCall): PolicyDecision
    {
        $name = strtolower($toolCall->getName());

        foreach (['read', 'get', 'list', 'search', 'find', 'show'] as $pattern) {
            if (str_contains($name, $pattern)) {
                return PolicyDecision::Allow;
            }
        }

        return PolicyDecision::AskUser;
    }
}
```

--------------------------------

### Initialize PostgreSQL vector store and index blog

Source: https://github.com/symfony/ai/blob/main/demo/README.md

Initializes the PostgreSQL vector store and indexes the blog content. Run these commands to set up the store for the chatbot's context.

```shell
symfony console ai:store:setup ai.store.postgres.symfony_blog
symfony console ai:store:index blog -vv
```

--------------------------------

### Install Symfony AI Store with Composer

Source: https://github.com/symfony/ai/blob/main/src/store/README.md

Use this command to install the core Store component via Composer. This is the required first step before using any of the store bridges.

```bash
composer require symfony/ai-store
```

--------------------------------

### Configure location in environment file

Source: https://github.com/symfony/ai/blob/main/docs/components/platform/vertexai.rst

Example environment file configuration for the location, recommending us-central1 and commenting out the global location.

```bash
# Recommended: Use a region with comprehensive model support
GOOGLE_CLOUD_LOCATION=us-central1

# Avoid: Global location has limited model availability
# GOOGLE_CLOUD_LOCATION=global
```

--------------------------------

### Example INSTRUCTIONS.md for agent instructions

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/creating-extensions.rst

A Markdown file that tells agents when to use your tools. It maps CLI commands to Mate tools and states benefits. Keep it short because every line costs context.

```markdown
## My Extension

Use the Mate tools instead of the CLI for better results:

| Instead of...              | Use                                                    |
|----------------------------|--------------------------------------------------------|
| `my-cli command`           | `vendor/bin/mate tools:call my-tool`                   |
| `my-cli search "term"`     | `vendor/bin/mate tools:call my-search --term="term"`   |

### Benefits
- Structured output that AI can parse
- Better error handling and context
- Integrated with project configuration
```

--------------------------------

### Install skills with vendor/bin/mate skills:install

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/skills.rst

Installs the skills of every enabled extension. The Composer plugin runs this automatically after composer install and composer update; run it manually to sync. Use --dry-run to preview what would be installed, rebuilt, or removed without writing anything.

```terminal
$ vendor/bin/mate skills:install
```

--------------------------------

### Create platform with API key (project-scoped)

Source: https://github.com/symfony/ai/blob/main/docs/components/platform/vertexai.rst

Create a platform using an API key with a project-scoped endpoint. Requires location and project ID.

```php
$platform = Factory::createPlatform(
    $_ENV['GOOGLE_CLOUD_LOCATION'],
    $_ENV['GOOGLE_CLOUD_PROJECT'],
    apiKey: $_ENV['GOOGLE_CLOUD_VERTEX_API_KEY'],
);
```

--------------------------------

### Install sqlite-vec extension

Source: https://github.com/symfony/ai/blob/main/docs/components/store/sqlite.rst

Install the sqlite-vec extension using the upstream install script. This drops a 'vec0.*' loadable file into the current directory.

```shell
curl -L https://github.com/asg017/sqlite-vec/releases/download/v0.1.9/install.sh | sh
```

--------------------------------

### Use Mate CLI tools and resources

Source: https://github.com/symfony/ai/blob/main/src/mate/README.md

Point your coding agent at the CLI by listing available tools, inspecting a tool's schema, calling a tool, and reading a resource. The generated mate/AGENT_INSTRUCTIONS.md provides guidance for agents.

```bash
vendor/bin/mate tools:list                          # list available tools
vendor/bin/mate tools:inspect symfony-profiler-list # show a tool's parameters/schema
vendor/bin/mate tools:call symfony-profiler-list --limit=1
vendor/bin/mate resources:read symfony-profiler://profile/<token>
```

--------------------------------

### Complete fast-track configuration

Source: https://github.com/symfony/ai/blob/main/docs/cookbook/ai-bundle-fast-track.rst

A complete fast-track configuration example combining platform, store, vectorizer, and indexer settings.

```yaml
# config/packages/ai.yaml
ai:
    platform:
        openai:
            api_key: '%env(OPENAI_API_KEY)%'

    store:
        chromadb:
            knowledge_base:
                collection: 'docs'

    vectorizer:
        embeddings:
            platform: 'ai.platform.openai'
            model: 'text-embedding-3-small'

    indexer:
        docs:
            loader: 'Symfony\AI\Store\Document\Loader\TextFileLoader'
            vectorizer: 'ai.vectorizer.embeddings'
            store: 'ai.store.chromadb.knowledge_base'
```

--------------------------------

### Install with Composer

Source: https://github.com/symfony/ai/blob/main/docs/components/platform/models-dev.rst

Install the models.dev bridge package using Composer.

```terminal
$ composer require symfony/ai-models-dev-platform
```

--------------------------------

### Run tools:list

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/troubleshooting.rst

Run this command to establish that the tools exist.

```terminal
vendor/bin/mate tools:list
```

--------------------------------

### Install MCP bundle with Composer

Source: https://github.com/symfony/ai/blob/main/docs/bundles/mcp-bundle.rst

Install the symfony/mcp-bundle package using Composer.

```terminal
$ composer require symfony/mcp-bundle
```

--------------------------------

### ChromaDB: Create store with client

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Creates a ChromaDB store by connecting to a local ChromaDB instance and instantiating the Store with a collection name.

```php
use Codewithkyrian\ChromaDB\Factory;
use Symfony\AI\Store\Bridge\ChromaDb\Store;

$client = (new Factory())->withHost('localhost')->withPort(8000)->connect();
$store = new Store($client, 'my_documents');
```

--------------------------------

### Run Mate tools: list, inspect, and call

Source: https://github.com/symfony/ai/blob/main/demo/README.md

List available Mate tools, inspect the symfony-ai-features tool's parameters and JSON input schema, and run it. Also shows how to call the symfony-profiler-list tool with a limit option.

```shell
vendor/bin/mate tools:list                            # what is available
vendor/bin/mate tools:inspect symfony-ai-features     # parameters and JSON input schema
vendor/bin/mate tools:call symfony-ai-features        # run it
vendor/bin/mate tools:call symfony-profiler-list --limit=1
```

--------------------------------

### Install and update dependencies

Source: https://github.com/symfony/ai/blob/main/src/platform/AGENTS.md

Install or update project dependencies using Composer.

```bash
composer install
composer update
```

--------------------------------

### Install ClickHouse store bridge

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the ClickHouse store bridge package.

```terminal
$ composer require symfony/ai-click-house-store
```

--------------------------------

### Create platform with API key (global endpoint)

Source: https://github.com/symfony/ai/blob/main/docs/components/platform/vertexai.rst

Create a platform using an API key with the global endpoint. Simplest setup, does not require the google/auth package.

```php
$platform = Factory::createPlatform(
    apiKey: $_ENV['GOOGLE_CLOUD_VERTEX_API_KEY'],
);
```

--------------------------------

### Setup MongoDB store collection and index

Source: https://github.com/symfony/ai/blob/main/docs/components/store/mongodb.rst

Creates the collection and Atlas Vector Search index. No-op if they already exist.

```php
$store->setup();
```

--------------------------------

### Install symfony/ai-s3vectors-store

Source: https://github.com/symfony/ai/blob/main/docs/components/store/s3-vectors.rst

Install the AWS S3 Vectors bridge package using Composer.

```terminal
$ composer require symfony/ai-s3vectors-store
```

--------------------------------

### Instantiate an Agent with Platform and Model

Source: https://github.com/symfony/ai/blob/main/docs/components/agent.rst

Create an Agent instance by passing a PlatformInterface (created via OpenAi Factory) and a model name. This is the first step to using the agent.

```php
use Symfony\AI\Agent\Agent;
use Symfony\AI\Platform\Bridge\OpenAi\Factory;

$platform = Factory::createPlatform($apiKey);
$model = 'gpt-4o-mini';

$agent = new Agent($platform, $model);
```

--------------------------------

### Install Deepgram bridge with Composer

Source: https://github.com/symfony/ai/blob/main/docs/components/platform/deepgram.rst

Install the Deepgram platform bridge for Symfony AI.

```bash
    $ composer require symfony/ai-deepgram-platform
```

--------------------------------

### Install symfony/ai-type-safe-platform

Source: https://github.com/symfony/ai/blob/main/docs/components/platform/typesafe.rst

Install the TypeSafe platform bridge for Symfony AI using Composer.

```terminal
$ composer require symfony/ai-type-safe-platform
```

--------------------------------

### Call Mate tools with parameters

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/commands.rst

Shows examples of the tools:call command, including passing parameters as options, using a bare boolean flag, and repeating options for variadic parameters. The default format is 'pretty'.

```terminal
# No parameters
$ vendor/bin/mate tools:call server-info

# Parameters as options
$ vendor/bin/mate tools:call monolog-search --term=error --level=error

# A boolean as a bare flag
$ vendor/bin/mate tools:call monolog-search --term="^GET" --regex

# A variadic parameter
$ vendor/bin/mate tools:call <tool-name> --tag=a --tag=b
```

--------------------------------

### Install symfony/ai-chat with Composer

Source: https://github.com/symfony/ai/blob/main/docs/components/chat.rst

Install the Symfony AI Chat component using Composer.

```terminal
$ composer require symfony/ai-chat
```

--------------------------------

### Install Symfony AI Platform via Composer

Source: https://github.com/symfony/ai/blob/main/docs/components/platform.rst

Install the symfony/ai-platform package using Composer.

```terminal
$ composer require symfony/ai-platform
```

--------------------------------

### Install Cloudflare Vectorize store bridge

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Install the Cloudflare Vectorize store bridge package.

```terminal
$ composer require symfony/ai-cloudflare-store
```

--------------------------------

### Weaviate: Install package

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Installs the Symfony AI Weaviate store package via Composer.

```terminal
$ composer require symfony/ai-weaviate-store
```

--------------------------------

### Milvus: Install package

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Installs the Symfony AI Milvus store package via Composer.

```terminal
$ composer require symfony/ai-milvus-store
```

--------------------------------

### Configuring Mate parameters and services in mate/config.php

Source: https://github.com/symfony/ai/blob/main/docs/components/mate.rst

Shows the structure of mate/config.php, a Symfony DI configuration file. Set parameters for Mate and its extensions, and register custom services. The example sets the invocation command, PHP version, and a log directory parameter, and enables autowiring and autoconfiguration for services.

```php
// mate/config.php
use Symfony\Component\DependencyInjection\Loader\Configurator\ContainerConfigurator;

return static function (ContainerConfigurator $container): void {
    $container->parameters()
        ->set('mate.invocation', 'vendor/bin/mate')
        ->set('mate.php_version', '8.3')
        ->set('ai_mate_monolog.log_dir', '%mate.root_dir%/var/log')
    ;

    $container->services()
        ->defaults()
            ->autowire()
            ->autoconfigure()

        // Register your custom services here
    ;
};
```

--------------------------------

### Postgres store installation command

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Composer command to install the Postgres store bridge package.

```terminal
$ composer require symfony/ai-postgres-store
```

--------------------------------

### Mocking Platform Responses with MockPlatformFactory

Source: https://github.com/symfony/ai/blob/main/docs/components/platform.rst

Demonstrates creating mock platforms using fixed strings, model-keyed maps, or closures for full control over responses.

```php
// 1. Fixed string - every call returns a TextResult
$platform = MockPlatformFactory::createPlatform('Mock result');
echo $platform->invoke('gpt-4o-mini', 'What is the capital of France?')->asText(); // "Mock result"

// 2. Map keyed by model name - per-model response
$platform = MockPlatformFactory::createPlatform([
    'gpt-4o-mini' => 'cheap answer',
    'gpt-4o' => 'expensive answer',
]);

// 3. Closure - full control, branch on the payload or options
$platform = MockPlatformFactory::createPlatform(
    fn ($model, $payload, $options) => "Echo: {$payload}"
);
```

--------------------------------

### Vektor store installation command

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Composer command to install the Vektor store bridge package.

```terminal
$ composer require symfony/ai-vektor-store
```

--------------------------------

### List skills

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/troubleshooting.rst

Run this command to list installed skills. A skill in state 'disabled' is not installed.

```terminal
vendor/bin/mate skills:list
```

--------------------------------

### Verify Mate tools with vendor/bin/mate tools:list

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/integration.rst

Run this command from a Claude Code session to verify that Mate is available and its tools are listed. Works out of the box after running 'mate init'.

```terminal
$ vendor/bin/mate tools:list
```

--------------------------------

### Show discovered capabilities

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/commands.rst

Display all discovered tools, resources, and resource templates. The examples show filtering by type and by extension package name.

```terminal
$ vendor/bin/mate debug:capabilities
```

```terminal
$ vendor/bin/mate debug:capabilities --type=tool
```

```terminal
$ vendor/bin/mate debug:capabilities --extension=symfony/ai-monolog-mate-extension
```

```terminal
$ vendor/bin/mate debug:capabilities --extension=_custom
```

--------------------------------

### Setup with additional field definitions

Source: https://github.com/symfony/ai/blob/main/docs/components/store/mongodb.rst

Passes additional field definitions to include extra fields in the search index.

```php
$store->setup([
    'fields' => [
        [
            'path' => 'title',
            'type' => 'string',
        ],
    ],
]);
```

--------------------------------

### Install symfony/ai-hugging-face-platform via Composer

Source: https://github.com/symfony/ai/blob/main/src/platform/src/Bridge/HuggingFace/README.md

Use this command to install the HuggingFace platform package for Symfony AI.

```bash
composer require symfony/ai-hugging-face-platform
```

--------------------------------

### Pinecone: Install packages

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Installs the Symfony AI Pinecone store package and the required probots-io/pinecone-php dependency.

```terminal
$ composer require symfony/ai-pinecone-store probots-io/pinecone-php
```

--------------------------------

### List Mate tools with filters and formats

Source: https://github.com/symfony/ai/blob/main/docs/components/mate/commands.rst

Shows examples of the tools:list command, including filtering by name pattern, extension package, and output format. The default format is 'table'.

```terminal
$ vendor/bin/mate tools:list
$ vendor/bin/mate tools:list --filter="monolog*"
$ vendor/bin/mate tools:list --extension=symfony/ai-monolog-mate-extension --filter="*search"
$ vendor/bin/mate tools:list --format=json
```

--------------------------------

### ChromaDB: Install packages

Source: https://github.com/symfony/ai/blob/main/docs/components/store/bridges.rst

Installs the Symfony AI ChromaDB store package and the required codewithkyrian/chromadb-php dependency.

```terminal
$ composer require symfony/ai-chroma-db-store codewithkyrian/chromadb-php
```

--------------------------------

### Install OpenMeteo Tool

Source: https://github.com/symfony/ai/blob/main/docs/components/agent/tools.rst

Install the OpenMeteo tool bridge via Composer. Requires the symfony/ai-open-meteo-tool package.

```terminal
$ composer require symfony/ai-open-meteo-tool
```

--------------------------------

### Calling tools and reading resources from the command line

Source: https://github.com/symfony/ai/blob/main/docs/components/mate.rst

Shows the command-line syntax for calling a tool with arguments and reading a resource by address. Use these commands to interact with Mate from a terminal or agent.

```terminal
$ vendor/bin/mate tools:call my-tool --param=value
$ vendor/bin/mate resources:read my-app://entity/42
```

--------------------------------

### Install Mapbox Tool

Source: https://github.com/symfony/ai/blob/main/docs/components/agent/tools.rst

Install the Mapbox tool bridge via Composer. Requires the symfony/ai-mapbox-tool package.

```terminal
$ composer require symfony/ai-mapbox-tool
```

--------------------------------

### Build a multi-provider platform with Platform and CatalogBasedModelRouter

Source: https://github.com/symfony/ai/blob/main/docs/components/platform.rst

Manually construct a Platform with multiple providers (OpenAI and Anthropic) to route invocations to the first provider whose catalog knows the requested model. The default CatalogBasedModelRouter handles routing.

```php
use Symfony\AI\Platform\Bridge\Anthropic\Factory as AnthropicFactory;
use Symfony\AI\Platform\Bridge\OpenAi\Factory as OpenAiFactory;
use Symfony\AI\Platform\Platform;

$platform = new Platform([
    OpenAiFactory::createProvider(apiKey: env('OPENAI_API_KEY')),
    AnthropicFactory::createProvider(apiKey: env('ANTHROPIC_API_KEY')),
]);

$platform->invoke('gpt-4o', $messages);             // → OpenAI
$platform->invoke('claude-3-5-sonnet', $messages);  // → Anthropic
```

--------------------------------

### Install Wikipedia Tool

Source: https://github.com/symfony/ai/blob/main/docs/components/agent/tools.rst

Install the Wikipedia tool bridge via Composer. Requires the symfony/ai-wikipedia-tool package.

```terminal
$ composer require symfony/ai-wikipedia-tool
```
Tokens:
Raw