Struct ClaudeCodeEngine 

pub struct ClaudeCodeEngine { /* private fields */ }

Implementations

impl ClaudeCodeEngine

pub fn new(config: ClaudeCodeConfig) -> Self

pub fn new_with_runner( config: ClaudeCodeConfig, runner: Arc<dyn ClaudeCodeRunner>, ) -> Self

Trait Implementations

impl InferenceEngine for ClaudeCodeEngine

fn id(&self) -> InferenceEngineId

fn capabilities(&self) -> InferenceCapabilities

fn metadata(&self) -> InferenceProviderMetadata

fn list_models<'life0, 'life1, 'async_trait>( &'life0 self, _ctx: InferenceProviderContext<'life1>, ) -> Pin<Box<dyn Future<Output = Result<Vec<ModelDescriptor>>> + Send + 'async_trait>>

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

fn stream_turn<'life0, 'life1, 'async_trait>( &'life0 self, ctx: InferenceTurnContext<'life1>, request: AgentInferenceRequest, ) -> Pin<Box<dyn Future<Output = Result<InferenceEventStream>> + Send + 'async_trait>>

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

fn tool_result_image_input(&self, _model: &str) -> bool

Whether this engine puts the image a tool result carries under crate::transcript::VIEW_IMAGE_DISPLAY_KEY in front of model.

fn requires_native_compaction(&self) -> bool

Whether compaction must remain provider-owned, without local pruning or text-summary fallback. Native failures propagate to the caller.

fn compact_turn<'life0, 'life1, 'async_trait>( &'life0 self, _ctx: InferenceTurnContext<'life1>, _request: AgentInferenceRequest, ) -> Pin<Box<dyn Future<Output = Result<Option<Pin<Box<dyn Stream<Item = Result<InferenceEvent, Error>> + Send>>>, Error>> + Send + 'async_trait>>

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

Native, opaque provider compaction. Unsupported engines return None. The stream publishes a candidate boundary and one terminal completion. Hosts must stage candidates until successful terminal completion; failed attempts may report billable usage but must not replace replay history.