Documentation
¶
Overview ¶
Package mediagen provides the built-in image__generate and video__generate tools. They let a text agent generate media as one step of a conversation by resolving an image/video provider from the shared provider pool (base.Registry) and invoking its Generate method.
The package lives in runtime so both the SDK and Arena can wire it without duplicating provider-resolution logic. It depends only on runtime/providers/base, runtime/tools, and runtime/types — never on sdk or tools/arena.
Index ¶
- Constants
- func ImageGenerateDescriptor() *tools.ToolDescriptor
- func VideoGenerateDescriptor() *tools.ToolDescriptor
- type Executor
- func (e *Executor) Execute(ctx context.Context, descriptor *tools.ToolDescriptor, args json.RawMessage) (json.RawMessage, error)
- func (e *Executor) ExecuteMultimodal(ctx context.Context, descriptor *tools.ToolDescriptor, args json.RawMessage) (json.RawMessage, []types.ContentPart, error)
- func (e *Executor) Name() string
Constants ¶
const ( // ImageGenerateToolName is the qualified name of the image generation tool. ImageGenerateToolName = "image__generate" // VideoGenerateToolName is the qualified name of the video generation tool. VideoGenerateToolName = "video__generate" // ExecutorMode is the tool Mode that routes to the mediagen Executor. ExecutorMode = "mediagen" )
Variables ¶
This section is empty.
Functions ¶
func ImageGenerateDescriptor ¶
func ImageGenerateDescriptor() *tools.ToolDescriptor
ImageGenerateDescriptor returns the descriptor for the image__generate tool.
func VideoGenerateDescriptor ¶
func VideoGenerateDescriptor() *tools.ToolDescriptor
VideoGenerateDescriptor returns the descriptor for the video__generate tool.
Types ¶
type Executor ¶
type Executor struct {
// contains filtered or unexported fields
}
Executor resolves an image/video provider from the pool and invokes its Generate method, returning the result as a media ContentPart that flows into the conversation message log. It implements tools.MultimodalExecutor.
func NewExecutor ¶
NewExecutor returns an Executor backed by the given provider pool. The pool may be nil; in that case every call resolves to a "no provider configured" error returned to the calling agent.
func (*Executor) Execute ¶
func (e *Executor) Execute( ctx context.Context, descriptor *tools.ToolDescriptor, args json.RawMessage, ) (json.RawMessage, error)
Execute satisfies tools.Executor. mediagen always produces media parts, so this returns the JSON summary and drops the parts; the registry prefers ExecuteMultimodal where available.
func (*Executor) ExecuteMultimodal ¶
func (e *Executor) ExecuteMultimodal( ctx context.Context, descriptor *tools.ToolDescriptor, args json.RawMessage, ) (json.RawMessage, []types.ContentPart, error)
ExecuteMultimodal generates media and returns a JSON summary plus the media ContentParts. An unresolved provider, a provider error, or empty output is returned as an error so the agent sees a clear tool failure rather than a silent no-op.
Source Files