Struct GenerateImageToolBox 

pub struct GenerateImageToolBox { /* private fields */ }

ToolBox that lets an agent generate and edit images with an ImageGenerator.

Tools:

• generate_image: draws an image from a prompt.
• edit_image: changes stored images the model names by ID. It needs a storage (with_storage) and a provider that can edit; without them the model gets an error.

Both take a name from the model (for example lighthouse), set as Media::name and so passed to MediaSave; the storage chooses the ID and may use it.

With a storage every image is saved and the model gets its ID. What else the model gets is set with with_output.

let storage = MemoryStorage::new();
let images = GenerateImageToolBox::new(ImageGenerator::new("openai::gpt-image-1")?)
    .with_storage(storage.clone());

Implementations

impl GenerateImageToolBox

pub fn new(generator: ImageGenerator) -> Self

Creates the toolbox. A storage set on generator is used to save the images, but edit_image needs one set with with_storage.

pub fn with_storage(self, storage: impl MediaStorage) -> Self

Saves every image in storage and loads the images to edit from it.

pub fn with_output(self, output: ImageToolOutput) -> Self

Sets what the model gets back, see ImageToolOutput.

Trait Implementations

impl ToolBox for GenerateImageToolBox

fn tools_definitions(&self) -> Result<Vec<Tool>, ToolError>

Returns a list of all Tool instances contained within this ToolBox. These definitions include the tool’s name, description, and parameters, which are used by the language model to decide which tool to call.

fn call_tool<'life0, 'life1, 'async_trait>( &'life0 self, call: &'life1 ToolCall, ) -> Pin<Box<dyn Future<Output = Result<ToolOutput, ToolError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Runs the tool call the model made.