### Agent Builder

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/struct.Agent.html

Demonstrates how to start building an Agent using the `builder` function.

```APIDOC
## impl<P: ProviderMarker> Agent<P>

### pub fn builder( client: impl Into<Arc<BoundClient<P>>>, model: impl IntoModelRef<P>, ) -> AgentBuilder<P>

Starts building an `Agent` from a provider-bound client and model.
```

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

### SSE Format Example

Source: https://docs.rs/aquaregia/latest/aquaregia/stream/index.html

Example of the text-based Server-Sent Events format used by the parser.

```text
event: message_type
data: {"json": "payload"}

data: [DONE]

```

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

### Quick Start: OpenAI Compatible Client

Source: https://docs.rs/aquaregia/latest/aquaregia/index.html

Demonstrates how to initialize an `LlmClient` for an OpenAI-compatible endpoint and generate text. Ensure the `DEEPSEEK_API_KEY` environment variable is set.

```rust
use aquaregia::{GenerateTextRequest, LlmClient};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let client = LlmClient::openai_compatible("https://api.deepseek.com")
        .api_key(std::env::var("DEEPSEEK_API_KEY")?)
        .build()?;

    let out = client
        .generate(GenerateTextRequest::from_user_prompt(
            "deepseek-chat",
            "Explain Rust ownership in 3 bullet points.",
        ))
        .await?;

    println!("{}", out.output_text);
    Ok(())
}
```

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

### Agent Builder Initialization

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/struct.Agent.html

Starts the process of building an Agent. Requires a provider-bound client and a model reference.

```rust
pub fn builder( client: impl Into<Arc<BoundClient<P>>>, model: impl IntoModelRef<P>, ) -> AgentBuilder<P>
```

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

### Generate Text with OpenAI Adapter

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai/index.html

Example of initializing the OpenAI client and generating text using a user prompt. Ensure you have the 'api-key' for authentication. The `build()` method returns a Result, so error handling is necessary.

```rust
use aquaregia::{LlmClient, GenerateTextRequest};

let client = LlmClient::openai("api-key").build()?;

let response = client
    .generate(GenerateTextRequest::from_user_prompt("gpt-4o", "Hello!"))
    .await?;

println!("{}", response.output_text);
```

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

### Example Usage

Source: https://docs.rs/aquaregia/latest/aquaregia/struct.CancellationToken.html

Demonstrates how to use CancellationToken to coordinate between tasks, allowing one task to signal cancellation to another.

```APIDOC
## Examples

### Basic Cancellation
```rust
use tokio::select;
use tokio_util::sync::CancellationToken;

#[tokio::main]
async fn main() {
    let token = CancellationToken::new();
    let cloned_token = token.clone();

    let join_handle = tokio::spawn(async move {
        // Wait for either cancellation or a very long time
        select! {
            _ = cloned_token.cancelled() => {
                // The token was cancelled
                5
            }
            _ = tokio::time::sleep(std::time::Duration::from_secs(9999)) => {
                99
            }
        }
    });

    tokio::spawn(async move {
        tokio::time::sleep(std::time::Duration::from_millis(10)).await;
        token.cancel();
    });

    assert_eq!(5, join_handle.await.unwrap());
}
```
```

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

### Anthropic Adapter Usage

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/anthropic/index.html

Example demonstrating how to initialize and use the Anthropic adapter for text generation.

```APIDOC
## POST /api/generate_text

### Description
Generates text using the Anthropic API via the Aquaregia adapter.

### Method
POST

### Endpoint
/api/generate_text

### Parameters
#### Query Parameters
- **model** (string) - Required - The name of the Anthropic model to use (e.g., "claude-sonnet-4-5").

#### Request Body
- **prompt** (string) - Required - The user's prompt for text generation.

### Request Example
```rust
use aquaregia::{LlmClient, GenerateTextRequest};

let client = LlmClient::anthropic("api-key").build()?;

let response = client
    .generate(GenerateTextRequest::from_user_prompt("claude-sonnet-4-5", "Hello!"))
    .await?;

println!("{}", response.output_text);
```

### Response
#### Success Response (200)
- **output_text** (string) - The generated text output from the Anthropic model.

#### Response Example
```json
{
  "output_text": "This is the generated text."
}
```
```

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

### Struct AgentStepStart

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.AgentStepStart.html

Represents the start of an agent step, including the step index and messages.

```APIDOC
## Struct AgentStepStart

### Description
Emitted when an agent step begins.

### Fields
- **step** (u8) - 1-based step index.
- **messages** (Vec<Message>) - Messages sent to the model for this step.

### Example
```rust
{
  "step": 1,
  "messages": [
    {
      "role": "user",
      "content": "Hello!"
    }
  ]
}
```
```

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

### Adjusting AgentRunPlan via prepare_call

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/struct.AgentRunPlan.html

Example of using the prepare_call callback to dynamically modify the AgentRunPlan, such as setting the temperature before execution.

```rust
let client = LlmClient::openai("key").build()?;

let agent = Agent::builder(client, "gpt-4o")
    .prepare_call(|plan| {
        // Dynamically adjust temperature based on task
        plan.temperature = Some(0.7);
    })
    .build()?;
```

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

### AgentStart Struct Definition

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.AgentStart.html

Defines the structure for agent start information, including model ID, messages, tool count, and max steps. This is emitted once before the first agent step.

```rust
pub struct AgentStart {
    pub model_id: String,
    pub messages: Vec<Message>,
    pub tool_count: usize,
    pub max_steps: u8,
}
```

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

### Initialize a tool descriptor with aquaregia::tool

Source: https://docs.rs/aquaregia/latest/aquaregia/tool/fn.tool.html

Use this function to start the builder pattern for defining a tool. It accepts any type that implements Into<String> for the tool name.

```rust
pub fn tool(name: impl Into<String>) -> ToolBuilder
```

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

### Build and Run an Agent with Tools

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/index.html

Demonstrates how to build an agent with a specific LLM client, define custom tools, and run the agent with a prompt. Ensure the LLM client is correctly initialized and the tools are properly defined with descriptions.

```rust
use aquaregia::{Agent, LlmClient, tool};
use serde_json::{Value, json};

#[tool(description = "Get weather by city")]
async fn get_weather(city: String) -> Result<Value, String> {
    Ok(json!({ "city": city, "temp_c": 23, "condition": "sunny" }))
}

let client = LlmClient::openai("api-key").build()?;

let agent = Agent::builder(client, "gpt-4o")
    .instructions("You can call tools before answering.")
    .tools([get_weather])
    .max_steps(4)
    .build()?;

let out = agent.run("What is the weather in Shanghai?").await?;
println!("{}", out.output_text);
```

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

### Pin::as_mut

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

Gets a mutable reference to the pinned value.

```APIDOC
## GET /pin/as_mut

### Description
Gets a mutable reference to the pinned value this `Pin<Ptr>` points to. This is a generic method to go from `&mut Pin<Pointer<T>>` to `Pin<&mut T>`. It is safe because, as part of the contract of `Pin::new_unchecked`, the pointee cannot move after `Pin<Pointer<T>>` got created. “Malicious” implementations of `Pointer::DerefMut` are likewise ruled out by the contract of `Pin::new_unchecked`. This method is useful when doing multiple calls to functions that consume the pinning pointer.

### Method
GET

### Endpoint
`/pin/as_mut`

### Parameters
#### Query Parameters
- **ptr** (Pin<Ptr>) - Required - The mutable pinned pointer to get a mutable reference from.

### Request Example
```rust
use std::pin::Pin;

// Assuming 'Type' is a struct with a method that takes Pin<&mut Self>
// impl Type {
//     fn method(self: Pin<&mut Self>) {
//         // do something
//     }
// }

// let mut instance: Type = ...;
// let mut pinned_instance: Pin<&mut Type> = Pin::new(&mut instance);
// pinned_instance.as_mut().method();
// pinned_instance.as_mut().method();
```

### Response
#### Success Response (200)
- **Pin<&mut T>** (Pin) - A mutable reference to the pinned value.
```

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

### Pin::as_ref

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

Gets a shared reference to the pinned value.

```APIDOC
## GET /pin/as_ref

### Description
Gets a shared reference to the pinned value this `Pin` points to. This is a generic method to go from `&Pin<Pointer<T>>` to `Pin<&T>`. It is safe because, as part of the contract of `Pin::new_unchecked`, the pointee cannot move after `Pin<Pointer<T>>` got created. “Malicious” implementations of `Pointer::Deref` are likewise ruled out by the contract of `Pin::new_unchecked`.

### Method
GET

### Endpoint
`/pin/as_ref`

### Parameters
#### Query Parameters
- **ptr** (Pin<Ptr>) - Required - The pinned pointer to get a shared reference from.

### Response
#### Success Response (200)
- **Pin<&T>** (Pin) - A shared reference to the pinned value.
```

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

### OpenAI-Compatible Adapter Initialization and Usage

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai_compatible/index.html

Demonstrates how to initialize the LlmClient with an OpenAI-compatible endpoint and make a text generation request.

```APIDOC
## POST /v1/chat/completions (Example Endpoint)

### Description
This example demonstrates initializing the `LlmClient` for an OpenAI-compatible endpoint and performing a text generation task.

### Method
POST (Implied by `generate` method)

### Endpoint
(Not explicitly defined, but typically `/v1/chat/completions` for OpenAI-compatible APIs)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
(Implicitly handled by `GenerateTextRequest`)
- **model** (string) - Required - The model to use for generation (e.g., "deepseek-chat").
- **messages** (array) - Required - The conversation history.
  - **role** (string) - Required - The role of the message sender (e.g., "user", "system", "assistant").
  - **content** (string) - Required - The content of the message.

### Request Example
```rust
use aquaregia::{LlmClient, GenerateTextRequest};

let client = LlmClient::openai_compatible("https://api.deepseek.com")
    .api_key("api-key")
    .build()?;

let response = client
    .generate(GenerateTextRequest::from_user_prompt("deepseek-chat", "Hello!"))
    .await?;

println!("{}", response.output_text);
```

### Response
#### Success Response (200)
- **output_text** (string) - The generated text output from the model.

#### Response Example
```json
{
  "output_text": "Hello there! How can I help you today?"
}
```
```

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

### Get Message Role

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.Message.html

Returns the role of the message sender.

```rust
pub fn role(&self) -> MessageRole
```

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

### Create New OpenAiCompatibleAdapterSettings

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai_compatible/struct.OpenAiCompatibleAdapterSettings.html

Initializes OpenAiCompatibleAdapterSettings with a base URL. Uses default path and no API key.

```rust
pub fn new(base_url: impl Into<String>) -> Self
```

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

### Get Message Name

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.Message.html

Returns the optional name attached to the message.

```rust
pub fn name(&self) -> Option<&str>
```

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

### Get Provider Kind

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.ModelRef.html

Returns the provider family marker for the model reference.

```rust
pub fn provider_kind(&self) -> ProviderKind
```

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

### Initialize and use an OpenAI-compatible LLM client

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai_compatible/index.html

Demonstrates how to configure an LlmClient with an OpenAI-compatible endpoint and perform a text generation request.

```rust
use aquaregia::{LlmClient, GenerateTextRequest};

let client = LlmClient::openai_compatible("https://api.deepseek.com")
    .api_key("api-key")
    .build()?;

let response = client
    .generate(GenerateTextRequest::from_user_prompt("deepseek-chat", "Hello!"))
    .await?;

println!("{}", response.output_text);
```

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

### Get Message Parts

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.Message.html

Returns a slice of the message's content parts.

```rust
pub fn parts(&self) -> &[ContentPart]
```

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

### Initialize Anthropic Client and Generate Text

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/anthropic/index.html

Demonstrates how to initialize the LlmClient for the Anthropic provider and generate text using a user prompt. Ensure you have the necessary API key and dependencies.

```rust
use aquaregia::{LlmClient, GenerateTextRequest};

let client = LlmClient::anthropic("api-key").build()?;

let response = client
    .generate(GenerateTextRequest::from_user_prompt("claude-sonnet-4-5", "Hello!"))
    .await?;

println!("{}", response.output_text);
```

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

### Client Initialization

Source: https://docs.rs/aquaregia/latest/aquaregia/client/index.html

Configuring and building a provider-bound client using the builder pattern.

```APIDOC
## Client Initialization

### Description
Initializes a client using LlmClient, configures runtime behavior via ClientBuilder, and produces a BoundClient.

### Parameters
#### Configuration
- **timeout** (Duration) - Optional - Sets the HTTP request timeout.
- **max_retries** (u32) - Optional - Sets the maximum number of retry attempts for failed requests.

### Request Example
let client = LlmClient::openai("api-key")
    .timeout(std::time::Duration::from_secs(60))
    .max_retries(3)
    .build()?;
```

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

### Pin::as_deref_mut

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

Gets `Pin<&mut T>` to the underlying pinned value from a nested `Pin`-pointer.

```APIDOC
## GET /pin/as_deref_mut

### Description
Gets `Pin<&mut T>` to the underlying pinned value from this nested `Pin`-pointer. This is a generic method to go from `Pin<&mut Pin<Pointer<T>>>` to `Pin<&mut T>`. It is safe because the existence of a `Pin<Pointer<T>>` ensures that the pointee, `T`, cannot move in the future, and this method does not enable the pointee to move. “Malicious” implementations of `Ptr::DerefMut` are likewise ruled out by the contract of `Pin::new_unchecked`.

### Method
GET

### Endpoint
`/pin/as_deref_mut`

### Parameters
#### Query Parameters
- **nested_pin** (Pin<&mut Pin<Ptr>>) - Required - The nested pinned pointer.

### Response
#### Success Response (200)
- **Pin<&mut T>** (Pin) - A mutable reference to the underlying pinned value.
```

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

### OpenAiAdapter Constructor

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai/struct.OpenAiAdapter.html

This snippet shows how to create an instance of the OpenAiAdapter using provided settings and an HTTP client.

```APIDOC
## POST /api/openai/adapter

### Description
Creates an adapter from validated settings and shared HTTP client.

### Method
POST

### Endpoint
/api/openai/adapter

### Parameters
#### Request Body
- **settings** (OpenAiAdapterSettings) - Required - The settings for the OpenAI adapter.
- **http** (Arc<Client>) - Required - A shared HTTP client instance.

### Request Example
```json
{
  "settings": {
    "api_key": "YOUR_API_KEY",
    "model": "gpt-3.5-turbo"
  },
  "http": "<Arc<Client> representation>"
}
```

### Response
#### Success Response (200)
- **OpenAiAdapter** (OpenAiAdapter) - The newly created OpenAI adapter instance.

#### Response Example
```json
{
  "adapter": "<OpenAiAdapter instance>"
}
```
```

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

### Get Provider Slug

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.ModelRef.html

Returns the static string slice representing the provider's slug.

```rust
pub fn provider_slug(&self) -> &'static str
```

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

### Create OpenAiAdapter from Settings

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai/struct.OpenAiAdapter.html

Constructs an OpenAiAdapter instance using provided settings and an HTTP client. Ensure settings are validated before use.

```rust
pub fn from_settings(settings: OpenAiAdapterSettings, http: Arc<Client>) -> Self>
```

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

### Get Agent Model ID

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/struct.Agent.html

Retrieves the fully qualified model identifier, formatted as '<provider>/<model>'.

```rust
pub fn model_id(&self) -> String
```

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

### Initialize LLM Clients with Aquaregia

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/index.html

Demonstrates how to instantiate various LLM provider clients using the LlmClient builder pattern.

```rust
use aquaregia::LlmClient;

// OpenAI adapter
let openai_client = LlmClient::openai("api-key").build()?;

// Anthropic adapter
let anthropic_client = LlmClient::anthropic("api-key").build()?;

// Google adapter
let google_client = LlmClient::google("api-key").build()?;

// OpenAI-compatible adapter (e.g., DeepSeek, local LLMs)
let compatible_client = LlmClient::openai_compatible("https://api.example.com")
    .api_key("api-key")
    .build()?;
```

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

### Get Type ID

Source: https://docs.rs/aquaregia/latest/aquaregia/client/struct.BoundClient.html

Retrieves the unique TypeId of the current type. Useful for runtime type identification.

```rust
fn type_id(&self) -> TypeId
```

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

### Initialize and Use Aquaregia Client

Source: https://docs.rs/aquaregia/latest/aquaregia/client/index.html

Demonstrates the builder pattern to configure an OpenAI client with timeouts and retries, followed by a text generation request.

```rust
use aquaregia::{GenerateTextRequest, LlmClient};

// Create and build client
let client = LlmClient::openai("api-key")
    .timeout(std::time::Duration::from_secs(60))
    .max_retries(3)
    .build()?;

// Use client for generation
let response = client
    .generate(GenerateTextRequest::from_user_prompt("gpt-4o", "Hello!"))
    .await?;

println!("{}", response.output_text);
```

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

### Create Google Client Builder

Source: https://docs.rs/aquaregia/latest/aquaregia/client/struct.LlmClient.html

Use this function to create a builder for a Google client. Requires an API key.

```rust
pub fn google(api_key: impl Into<String>) -> ClientBuilder<Google>
```

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

### Generate text with GoogleAdapter

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/google/index.html

Initializes the Google client and performs a text generation request using the Gemini 2.0 Flash model.

```rust
use aquaregia::{LlmClient, GenerateTextRequest};

let client = LlmClient::google("api-key").build()?;

let response = client
    .generate(GenerateTextRequest::from_user_prompt("gemini-2.0-flash", "Hello!"))
    .await?;

println!("{}", response.output_text);
```

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

### Get Canonical Provider Slug

Source: https://docs.rs/aquaregia/latest/aquaregia/types/enum.ProviderKind.html

Returns the normalized string identifier for the provider, suitable for use in URLs, logging, or configuration.

```rust
pub fn as_slug(&self) -> &'static str
```

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

### Create OpenAI-Compatible Client Builder

Source: https://docs.rs/aquaregia/latest/aquaregia/client/struct.LlmClient.html

Use this function to create a builder for an OpenAI-compatible client. Requires a base URL.

```rust
pub fn openai_compatible( base_url: impl Into<String>, ) -> ClientBuilder<OpenAiCompatible>
```

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

### Implement ToString for T

Source: https://docs.rs/aquaregia/latest/aquaregia/tool/enum.ToolExecError.html

Provides a method to convert any type implementing the Display trait into a String. This is a convenient way to get a string representation.

```rust
fn to_string(&self) -> String
```

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

### Create new AnthropicAdapterSettings

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/anthropic/struct.AnthropicAdapterSettings.html

Constructor for initializing settings with a provided API key and default values for base URL and API version.

```rust
pub fn new(api_key: impl Into<String>) -> Self
```

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

### Get All Registered Tool Names

Source: https://docs.rs/aquaregia/latest/aquaregia/tool/struct.ToolRegistry.html

Returns a vector containing the names of all tools currently registered in the ToolRegistry. The names are returned as string slices.

```rust
pub fn names(&self) -> Vec<&str>
```

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

### Get Tool by Name

Source: https://docs.rs/aquaregia/latest/aquaregia/tool/struct.ToolRegistry.html

Retrieves a registered tool from the ToolRegistry by its name. Returns an Option<&Tool>, which is Some(&Tool) if the tool is found, or None otherwise.

```rust
pub fn get(&self, name: &str) -> Option<&Tool>
```

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

### Get Provider-Local Model ID

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.ModelRef.html

Retrieves the model identifier as used by the specific provider, without the provider prefix. This is typically used in API requests.

```rust
pub fn model(&self) -> &str
```

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

### AgentStart Data Structure

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.AgentStart.html

Definition and fields for the AgentStart object.

```APIDOC
## AgentStart Structure

### Description
Emitted once before the first agent step to initialize the run context.

### Fields
- **model_id** (String) - Fully-qualified model id used by the run.
- **messages** (Vec<Message>) - Initial messages passed into the run.
- **tool_count** (usize) - Number of registered tools.
- **max_steps** (u8) - Effective max step cap for this run.

### Request Example
{
  "model_id": "gpt-4-turbo",
  "messages": [],
  "tool_count": 5,
  "max_steps": 10
}
```

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

### PolicyExt::or

Source: https://docs.rs/aquaregia/latest/aquaregia/struct.CancellationToken.html

Creates a new Policy that returns Action::Follow if either policy returns Action::Follow.

```APIDOC
## POST /policy/or

### Description
Create a new `Policy` that returns `Action::Follow` if either `self` or `other` returns `Action::Follow`.

### Method
POST

### Endpoint
/policy/or

### Parameters
#### Request Body
- **other** (Policy) - Required - The other policy to combine.
- **self** (Policy) - Required - The first policy.

### Request Example
```json
{
  "other": "<other_policy_representation>",
  "self": "<self_policy_representation>"
}
```

### Response
#### Success Response (200)
- **Or<T, P>** - A new policy representing the logical OR of the two input policies.

#### Response Example
```json
{
  "policy": "<or_policy_representation>"
}
```
```

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

### Get Fully-Qualified Model ID

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.ModelRef.html

Retrieves the complete model identifier, including the provider prefix (e.g., "provider/model"). This is useful for logging or display purposes.

```rust
pub fn id(&self) -> String
```

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

### Unwrapping a Pin to Get the Underlying Pointer

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextDeltaStream.html

Safely unwraps a `Pin<Ptr>` to return the original pointer `Ptr`. This operation is `unsafe` and should only be used when the pinning invariants are guaranteed to be upheld.

```rust
pub const unsafe fn into_inner_unchecked(pin: Pin<Ptr>) -> Ptr
```

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

### Create OpenAI Client Builder

Source: https://docs.rs/aquaregia/latest/aquaregia/client/struct.LlmClient.html

Use this function to create a builder for an OpenAI client. Requires an API key.

```rust
pub fn openai(api_key: impl Into<String>) -> ClientBuilder<OpenAi>
```

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

### Getting Nested Mutable Reference from Pin

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextDeltaStream.html

Extracts a `Pin<&mut T>` from a nested `Pin<&mut Pin<Pointer<T>>>`. This operation is safe because the outer `Pin` guarantees the inner pointee cannot move.

```rust
pub fn as_deref_mut( self: Pin<&mut Pin<Ptr>>, ) -> Pin<&mut <Ptr as Deref>::Target> where Ptr: DerefMut
```

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

### OpenAiAdapterSettings Configuration

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai/struct.OpenAiAdapterSettings.html

Defines the structure and initialization for OpenAI adapter settings.

```APIDOC
## OpenAiAdapterSettings

### Description
Runtime settings for the OpenAI adapter, containing configuration for the API endpoint and authentication.

### Fields
- **base_url** (String) - Base URL for API requests.
- **api_key** (String) - API key sent as a bearer token.

### Initialization
#### pub fn new(api_key: impl Into<String>) -> Self
Creates settings with a default base URL.
```

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

### Getting a Shared Reference to Pinned Data

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextDeltaStream.html

Retrieves a shared reference (`Pin<&T>`) to the data pointed to by a `Pin<Pointer<T>>`. This operation is safe due to the pinning contract ensuring the data does not move.

```rust
pub fn as_ref(&self) -> Pin<&<Ptr as Deref>::Target> where Ptr: Deref
```

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

### GoogleAdapterSettings Configuration

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/google/struct.GoogleAdapterSettings.html

Defines the structure and initialization for Google adapter settings.

```APIDOC
## GoogleAdapterSettings

### Description
Runtime settings for the Google adapter, used to configure API requests.

### Fields
- **base_url** (String) - Base URL for API requests.
- **api_key** (String) - API key sent via the x-goog-api-key header.

### Initialization
- **new(api_key: impl Into<String>) -> Self** - Creates settings with a default base URL.
```

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

### Provider-Specific Settings

Source: https://docs.rs/aquaregia/latest/aquaregia/client/trait.ProviderBinding.html

Illustrates the different configuration options for various AI providers. Each provider requires specific settings such as base URL and API key.

```rust
OpenAI: `base_url`, `api_key`
Anthropic: `base_url`, `api_key`, `api_version`
Google: `base_url`, `api_key`
OpenAI-compatible: `base_url`, optional `api_key`, custom headers/query params
```

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

### Formatting

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

Implementation of the `fmt::Display` trait for Pin.

```APIDOC
## Formatting

### `fmt`

- **Description**: Formats the value using the given formatter.
- **Method**: N/A (Method on `Pin<T>`)
- **Parameters**: `f: &mut Formatter<'_>`
- **Returns**: `Result<(), Error>`
```

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

### Getting a Mutable Reference to Pinned Data

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextDeltaStream.html

Retrieves a mutable reference (`Pin<&mut T>`) to the data pointed to by a `Pin<Pointer<T>>`. This is useful when multiple operations need to be performed on the pinned data, and each operation requires a mutable borrow.

```rust
pub fn as_mut(&mut self) -> Pin<&mut <Ptr as Deref>::Target> where Ptr: DerefMut
```

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

### Demonstrate UB with `Pin::new_unchecked` on `&mut T`

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

This example shows how calling `Pin::new_unchecked` on a mutable reference can lead to undefined behavior if the underlying data is moved after pinning. This violates the API contract that pinned data must not be moved.

```rust
use std::mem;
use std::pin::Pin;

fn move_pinned_ref<T>(mut a: T, mut b: T) {
    unsafe {
        let p: Pin<&mut T> = Pin::new_unchecked(&mut a);
        // This should mean the pointee `a` can never move again.
    }
    mem::swap(&mut a, &mut b); // Potential UB down the road ⚠️
    // The address of `a` changed to `b`'s stack slot, so `a` got moved even
    // though we have previously pinned it! We have violated the pinning API contract.
}
```

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

### OpenAiCompatibleAdapterSettings Configuration

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai_compatible/struct.OpenAiCompatibleAdapterSettings.html

Methods for configuring the OpenAI-compatible adapter settings, including base URL, authentication, and parsing options.

```APIDOC
## Configuration Methods

### Description
Methods to configure the behavior of the OpenAI-compatible adapter.

### Methods
- **new(base_url)**: Creates settings with a default path and no API key.
- **api_key(api_key)**: Sets a bearer token for authentication.
- **no_api_key()**: Clears the bearer token.
- **header(name, value)**: Adds or replaces a custom HTTP header.
- **query_param(name, value)**: Adds or replaces a query parameter.
- **chat_completions_path(path)**: Overrides the default chat completions path.
- **think_tag_parsing(enabled)**: Enables or disables <think>/<thinking> tag parsing from content.
- **think_tag_case_insensitive(case_insensitive)**: Controls case sensitivity for think tag parsing.
```

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

### Agent Run Methods

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/struct.Agent.html

Details the different methods for running the agent, including single prompt, message list, and cancellable operations.

```APIDOC
### pub async fn run( &self, prompt: impl Into<String>, ) -> Result<AgentResponse, Error>

Runs the agent with a single user prompt.
If `instructions` were configured, they are inserted as an initial system message.

### pub async fn run_messages( &self, messages: Vec<Message>, ) -> Result<AgentResponse, Error>

Runs the agent with an explicit message list.

### pub async fn run_cancellable( &self, prompt: impl Into<String>, token: CancellationToken, ) -> Result<AgentResponse, Error>

Runs the agent with a prompt and cancellation support.

### pub async fn run_messages_cancellable( &self, messages: Vec<Message>, token: CancellationToken, ) -> Result<AgentResponse, Error>

Runs the agent with explicit messages and cancellation support.
```

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

### Demonstrate UB with `Pin::new_unchecked` on `Rc<T>`

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

This example illustrates undefined behavior when using `Pin::new_unchecked` with `Rc<T>`. If other references to the same data exist, they are not subject to pinning restrictions, allowing the data to be moved while still considered pinned, thus violating the API contract.

```rust
use std::rc::Rc;
use std::pin::Pin;

fn move_pinned_rc<T>(mut x: Rc<T>) {
    // This should mean the pointee can never move again.
    let pin = unsafe { Pin::new_unchecked(Rc::clone(&x)) };
    {
        let p: Pin<&T> = pin.as_ref();
        // ...
    }
    drop(pin);

    let content = Rc::get_mut(&mut x).unwrap(); // Potential UB down the road ⚠️
    // Now, if `x` was the only reference, we have a mutable reference to
    // data that we pinned above, which we could use to move it as we have
    // seen in the previous example. We have violated the pinning API contract.
}
```

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

### Policy::or

Source: https://docs.rs/aquaregia/latest/aquaregia/client/struct.ClientBuilder.html

Creates a new Policy that returns Action::Follow if either the current or the provided policy returns Action::Follow.

```APIDOC
## fn or<P, B, E>(self, other: P) -> Or<T, P>

### Description
Create a new `Policy` that returns `Action::Follow` if either `self` or `other` returns `Action::Follow`.

### Parameters
#### Path Parameters
- **other** (P) - Required - The other policy to evaluate.
```

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

### Clone Implementation for AgentStart

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.AgentStart.html

Provides methods to duplicate or copy-assign AgentStart values. Ensures that AgentStart instances can be cloned.

```rust
fn clone(&self) -> AgentStart
```

```rust
fn clone_from(&mut self, source: &Self)
```

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

### Build BoundClient

Source: https://docs.rs/aquaregia/latest/aquaregia/client/struct.ClientBuilder.html

Builds a provider-bound client with the configured settings. This method validates the settings before returning the client.

```rust
pub fn build(self) -> Result<BoundClient<P>, Error>
```

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

### Run Agent with Prompt

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/struct.Agent.html

Executes the agent with a single user prompt. If configured, instructions are prepended as a system message.

```rust
pub async fn run( &self, prompt: impl Into<String>, ) -> Result<AgentResponse, Error>
```

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

### Clone Implementation for AgentPreparedStep

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.AgentPreparedStep.html

Provides the `clone` and `clone_from` methods for the AgentPreparedStep struct, allowing for duplication of its state. Requires the generic type P to implement Clone and ProviderMarker.

```rust
impl<P: Clone + ProviderMarker> Clone for AgentPreparedStep<P>
```

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

### Create AnthropicAdapter from Settings

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/anthropic/struct.AnthropicAdapter.html

Constructs an AnthropicAdapter instance using provided settings and a shared HTTP client. Ensure settings are validated before use.

```rust
pub fn from_settings( settings: AnthropicAdapterSettings, http: Arc<Client>, ) -> Self
```

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

### DispatchFromDyn Implementation

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

Implementation of `DispatchFromDyn` for `Pin<Ptr>` to `Pin<U>`.

```APIDOC
## DispatchFromDyn Implementation

### `DispatchFromDyn<Pin<U>>`

- **Description**: Allows dispatching from `Pin<Ptr>` to `Pin<U>` under specific trait bounds.
- **Method**: N/A (Implementation for `Pin<Ptr>`)
- **Bounds**: `Ptr: DispatchFromDyn<U> + PinCoerceUnsized`, `U: PinCoerceUnsized`
```

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

### Implement ProviderBinding for OpenAi

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.OpenAi.html

Binds the `OpenAi` provider with its specific settings and adapter. Use `into_adapter` to convert settings into a runtime adapter.

```rust
type Settings = OpenAiAdapterSettings
```

```rust
fn into_adapter(
    settings: Self::Settings,
    http: Arc<Client>,
) -> Arc<dyn ModelAdapter<Self>>
```

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

### Clone Implementation for Pin

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

Details the `clone` and `clone_from` methods for `Pin<Ptr>` when `Ptr` implements `Clone`.

```APIDOC
## impl<Ptr> Clone for Pin<Ptr>
where Ptr: Clone,

### fn clone(&self) -> Pin<Ptr>
Returns a duplicate of the value. Read more

### fn clone_from(&mut self, source: &Self)
Performs copy-assignment from `source`. Read more
```

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

### AsyncSeek Implementation for Pin

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

Details the `start_seek` and `poll_complete` methods for `Pin<P>` when `P` implements `AsyncSeek`.

```APIDOC
## impl<P> AsyncSeek for Pin<P>
where P: DerefMut, <P as Deref>::Target: AsyncSeek,

### fn start_seek(self: Pin<&mut Pin<P>>, pos: SeekFrom) -> Result<(), Error>
Attempts to seek to an offset, in bytes, in a stream. Read more

### fn poll_complete( self: Pin<&mut Pin<P>>, cx: &mut Context<'_>, ) -> Poll<Result<u64, Error>>
Waits for a seek operation to complete. Read more
```

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

### TryFrom Implementation

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/struct.AgentBuilder.html

Provides a mechanism to attempt conversion from one type to another, with a potential error.

```APIDOC
### impl<T, U> TryFrom<U> for T

#### type Error = Infallible

### Description
The type returned in the event of a conversion error.

#### fn try_from(value: U) -> Result<T, <T as TryFrom<U>>::Error>

### Description
Performs the conversion.

### Method
try_from

### Endpoint
N/A
```

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

### Tool::from_parts

Source: https://docs.rs/aquaregia/latest/aquaregia/tool/struct.Tool.html

Constructor for creating a Tool instance from its components.

```APIDOC
## impl Tool

### pub fn from_parts( descriptor: ToolDescriptor, executor: Arc<dyn ToolExecutor>, ) -> Self

Creates a tool from explicit parts.
```

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

### AgentPreparedStep Struct

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.AgentPreparedStep.html

Represents a finalized step input for an agent, containing model, messages, tools, and generation parameters.

```APIDOC
## Struct AgentPreparedStep

### Description
Finalized step input returned by `prepare_step`.

### Fields
- **model** (ModelRef<P>) - Model selected for this step.
- **messages** (Vec<Message>) - Messages to send for this step.
- **tools** (Vec<Tool>) - Tools available for this step.
- **temperature** (Option<f32>) - Sampling temperature for this step.
- **max_output_tokens** (Option<u32>) - Max output token budget for this step.
- **stop_sequences** (Vec<String>) - Stop sequences for this step.

### Trait Implementations
#### Clone
- `clone(&self) -> AgentPreparedStep<P>`: Returns a duplicate of the value.
- `clone_from(&mut self, source: &Self)`: Performs copy-assignment from `source`.

#### Debug
- `fmt(&self, f: &mut Formatter<'_>) -> Result`: Formats the value using the given formatter.

### Auto Trait Implementations
- Freeze
- !RefUnwindSafe
- Send
- Sync
- Unpin
- UnsafeUnpin
- !UnwindSafe

### Blanket Implementations
- Any
- Borrow<T>
- BorrowMut<T>
- CloneToUninit
- DynClone
- From<T>
- Instrument
- Into<U>
- PolicyExt
- ToOwned
- TryFrom<U>
- TryInto<U>
- WithSubscriber
```

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

### POST /stream

Source: https://docs.rs/aquaregia/latest/aquaregia/client/struct.BoundClient.html

Runs a streaming generation request using the bound provider.

```APIDOC
## POST /stream

### Description
Runs a streaming generation request. The request is validated locally and retried on retryable failures.

### Method
POST

### Parameters
#### Request Body
- **req** (GenerateTextRequest<P>) - Required - The generation request configuration.

### Response
#### Success Response (200)
- **stream** (TextStream) - A stream of text responses.
```

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

### TryInto Implementation

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/struct.AgentBuilder.html

Provides a mechanism to attempt conversion into another type, with a potential error.

```APIDOC
### impl<T, U> TryInto<U> for T

#### type Error = <U as TryFrom<T>>::Error

### Description
The type returned in the event of a conversion error.

#### fn try_into(self) -> Result<U, <U as TryFrom<T>>::Error>

### Description
Performs the conversion.

### Method
try_into

### Endpoint
N/A
```

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

### Clone Into AgentPrepareStep

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.AgentPrepareStep.html

Uses borrowed data to replace owned data, usually by cloning. Requires `T` to implement `Clone`.

```rust
fn clone_into(&self, target: &mut T)
```

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

### Create Anthropic Client Builder

Source: https://docs.rs/aquaregia/latest/aquaregia/client/struct.LlmClient.html

Use this function to create a builder for an Anthropic client. Requires an API key.

```rust
pub fn anthropic(api_key: impl Into<String>) -> ClientBuilder<Anthropic>
```

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

### Function: tool

Source: https://docs.rs/aquaregia/latest/aquaregia/tool/fn.tool.html

Initializes a ToolBuilder for defining tool descriptors and executors.

```APIDOC
## tool

### Description
Starts building a tool descriptor and executor.

### Signature
`pub fn tool(name: impl Into<String>) -> ToolBuilder`

### Parameters
- **name** (impl Into<String>) - Required - The name of the tool to be created.
```

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

### Define GoogleAdapterSettings struct

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/google/struct.GoogleAdapterSettings.html

The structure used to hold runtime settings for the Google adapter.

```rust
pub struct GoogleAdapterSettings {
    pub base_url: String,
    pub api_key: String,
}
```

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

### Combine Policies with And

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/anthropic/struct.AnthropicAdapter.html

Creates a new `Policy` that requires both `self` and `other` policies to return `Action::Follow`. Used for combining policy logic.

```rust
fn and<P, B, E>(self, other: P) -> And<T, P>
where T: Policy<B, E>,
 P: Policy<B, E>,
```

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

### Agent Runtime and Builder API

Source: https://docs.rs/aquaregia/latest/aquaregia/agent/index.html

Overview of the Agent architecture and usage for creating and executing tool-using agents.

```APIDOC
## Agent Runtime and Builder

### Description
The Agent module provides the core abstraction for multi-step tool-using agents. It manages the interaction loop between the LLM and available tools.

### Components
- **Agent<P>**: The main runtime for executing tool loops.
- **AgentBuilder<P>**: Used to configure agent behavior, including instructions, tools, and model parameters.
- **AgentRunPlan<P>**: A mutable plan used to customize specific agent runs.

### Usage Example
```rust
use aquaregia::{Agent, LlmClient, tool};
use serde_json::{Value, json};

#[tool(description = "Get weather by city")]
async fn get_weather(city: String) -> Result<Value, String> {
    Ok(json!({ "city": city, "temp_c": 23, "condition": "sunny" }))
}

let client = LlmClient::openai("api-key").build()?;

let agent = Agent::builder(client, "gpt-4o")
    .instructions("You can call tools before answering.")
    .tools([get_weather])
    .max_steps(4)
    .build()?;

let out = agent.run("What is the weather in Shanghai?").await?;
println!("{}", out.output_text);
```
```

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

### Conversion Traits (Into, TryFrom, TryInto)

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.ModelRef.html

Documentation for type conversion traits.

```APIDOC
### impl<T, U> Into<U> for T
where U: From<T>,

#### fn into(self) -> U

### Description
Calls `U::from(self)`. That is, this conversion is whatever the implementation of `From<T> for U` chooses to do.

### Method
`into`

### Endpoint
N/A (Method Implementation)

### impl<T, U> TryFrom<U> for T
where U: Into<T>,

#### type Error = Infallible

### Description
The type returned in the event of a conversion error.

#### fn try_from(value: U) -> Result<T, <T as TryFrom<U>>::Error>

### Description
Performs the conversion.

### Method
`try_from`

### Endpoint
N/A (Method Implementation)

### impl<T, U> TryInto<U> for T
where U: TryFrom<T>,

#### type Error = <U as TryFrom<T>>::Error

### Description
The type returned in the event of a conversion error.

#### fn try_into(self) -> Result<U, <U as TryFrom<T>>::Error>

### Description
Performs the conversion.

### Method
`try_into`

### Endpoint
N/A (Method Implementation)
```

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

### Implement ProviderMarker for Google

Source: https://docs.rs/aquaregia/latest/aquaregia/types/trait.ProviderMarker.html

Implementation of the ProviderMarker trait for the Google provider.

```rust
impl ProviderMarker for Google {
    const KIND: ProviderKind = ProviderKind::Google
}
```

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

### Policy AND Operation

Source: https://docs.rs/aquaregia/latest/aquaregia/model_adapters/openai_compatible/struct.OpenAiCompatibleAdapterSettings.html

Creates a new `Policy` that returns `Action::Follow` only if both `self` and `other` policies return `Action::Follow`. Part of `PolicyExt`.

```rust
fn and<P, B, E>(self, other: P) -> And<T, P>
where T: Policy<B, E>, P: Policy<B, E>
```

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

### Define a tool using the #[tool] macro

Source: https://docs.rs/aquaregia/latest/aquaregia/tool/index.html

The recommended approach for defining tools using the procedural macro for concise syntax.

```rust
use aquaregia::tool;
use serde_json::{Value, json};

#[tool(description = "Get weather by city")]
async fn get_weather(city: String) -> Result<Value, String> {
    Ok(json!({ "city": city, "temp_c": 23, "condition": "sunny" }))
}
```

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

### Comparison Methods

Source: https://docs.rs/aquaregia/latest/aquaregia/types/type.TextStream.html

Methods for comparing Pin instances.

```APIDOC
## Comparison Methods

### `partial_cmp`

- **Description**: Returns an ordering between `self` and `other` values if one exists.
- **Method**: N/A (Method on `Pin<T>`)
- **Parameters**: `other: &Pin<Q>`
- **Returns**: `Option<Ordering>`

### `lt`

- **Description**: Tests less than (for `self` and `other`) and is used by the `<` operator.
- **Method**: N/A (Method on `Pin<T>`)
- **Parameters**: `other: &Pin<Q>`
- **Returns**: `bool`

### `le`

- **Description**: Tests less than or equal to (for `self` and `other`) and is used by the `<=` operator.
- **Method**: N/A (Method on `Pin<T>`)
- **Parameters**: `other: &Pin<Q>`
- **Returns**: `bool`

### `gt`

- **Description**: Tests greater than (for `self` and `other`) and is used by the `>` operator.
- **Method**: N/A (Method on `Pin<T>`)
- **Parameters**: `other: &Pin<Q>`
- **Returns**: `bool`

### `ge`

- **Description**: Tests greater than or equal to (for `self` and `other`) and is used by the `>=` operator.
- **Method**: N/A (Method on `Pin<T>`)
- **Parameters**: `other: &Pin<Q>`
- **Returns**: `bool`
```

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

### AgentPrepareStep Struct Definition

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.AgentPrepareStep.html

Defines the structure for per-step mutable input passed to `prepare_step`. It includes fields for step details, model, messages, tools, and generation parameters.

```rust
pub struct AgentPrepareStep<P: ProviderMarker> {
    pub step: u8,
    pub model: ModelRef<P>,
    pub messages: Vec<Message>,
    pub tools: Vec<Tool>,
    pub temperature: Option<f32>,
    pub max_output_tokens: Option<u32>,
    pub stop_sequences: Vec<String>,
    pub previous_steps: Vec<AgentStep>,
}
```

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

### Usage Struct Methods

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.Usage.html

Methods for creating and modifying Usage instances.

```APIDOC
## impl Usage

### `from_totals`

#### Signature
`pub fn from_totals(input_tokens: u32, output_tokens: u32, reasoning_tokens: u32, total_tokens: Option<u32>) -> Self`

### Description
Builds usage from provider totals and back-fills derived counters.
```

```APIDOC
### `with_input_cache_split`

#### Signature
`pub fn with_input_cache_split(self, cache_read_tokens: u32, cache_write_tokens: u32) -> Self`

### Description
Sets input cache split and recomputes no-cache input tokens.
```

```APIDOC
### `with_output_split`

#### Signature
`pub fn with_output_split(self, output_text_tokens: u32, reasoning_tokens: u32) -> Self`

### Description
Sets output text/reasoning split and recomputes total output tokens.
```

```APIDOC
### `with_raw_usage`

#### Signature
`pub fn with_raw_usage(self, raw_usage: Value) -> Self`

### Description
Attaches raw provider usage payload.
```

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

### Create a User Message with Text

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.Message.html

Creates a user message containing a single text part. User messages represent prompts or questions from the end user.

```rust
pub fn user_text(text: impl Into<String>) -> Self
```

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

### Create ModelRef Instance

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.ModelRef.html

Instantiates a ModelRef for a specific provider. Requires importing the provider type (e.g., `OpenAi`) and the `ModelRef` struct. The model ID is a String.

```rust
use aquaregia::{ModelRef, OpenAi};

let model = ModelRef::<OpenAi>::new("gpt-4o");
assert_eq!(model.id(), "openai/gpt-4o");
assert_eq!(model.model(), "gpt-4o");
```

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

### Policy::or

Source: https://docs.rs/aquaregia/latest/aquaregia/types/enum.ProviderKind.html

Combines two policies into a new policy that returns Action::Follow if either policy succeeds.

```APIDOC
## fn or<P, B, E>(self, other: P) -> Or<T, P>

### Description
Creates a new `Policy` that returns `Action::Follow` if either `self` or `other` returns `Action::Follow`.

### Parameters
#### Path Parameters
- **other** (P) - Required - The secondary policy to evaluate.
```

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

### Create User Message with Text and Image URL

Source: https://docs.rs/aquaregia/latest/aquaregia/types/struct.Message.html

Creates a user message that includes both text and an image URL.

```rust
pub fn user_text_and_image_url( text: impl Into<String>, url: impl Into<String>, ) -> Self
```