Struct TapedLlm 

pub struct TapedLlm<L = Arc<dyn BaseLlm>> { /* private fields */ }

A BaseLlm that records its calls to a Tape or answers them from one. See the module docs.

Implementations

impl<L> TapedLlm<L>

where L: BaseLlm,

pub fn recording(inner: L, tape: Arc<dyn Tape>) -> TapedLlm<L>

Pass every call through to inner and record it on tape.

impl TapedLlm

pub fn replaying(model_id: impl Into<String>, tape: Arc<dyn Tape>) -> TapedLlm

Answer every call from tape. No model, network or credential is used; a call the tape does not hold fails with LlmError::Config.

Trait Implementations

impl<L> BaseLlm for TapedLlm<L>

where L: BaseLlm,

fn model_id(&self) -> &str

The model identifier (e.g., “gemini-2.5-flash”).

fn capabilities(&self) -> ModelCapabilities

What this model supports — the ADK model-capability-declaration pattern: the model states its capabilities instead of every caller re-inferring them from the id string. The default derives a conservative estimate from model_id; back-ends with authoritative knowledge should override.

fn generate<'life0, 'async_trait>( &'life0 self, request: LlmRequest, ) -> Pin<Box<dyn Future<Output = Result<LlmResponse, LlmError>> + Send + 'async_trait>>

where 'life0: 'async_trait, TapedLlm<L>: 'async_trait,

Generate content from the LLM.

fn warm_up<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = Result<(), LlmError>> + Send + 'async_trait>>

where 'life0: 'async_trait, TapedLlm<L>: 'async_trait,

Pre-warm the HTTP connection pool to avoid cold-start latency.

fn generate_stream<'life0, 'async_trait>( &'life0 self, request: LlmRequest, ) -> Pin<Box<dyn Future<Output = Result<Pin<Box<dyn Stream<Item = Result<LlmResponse, LlmError>> + Send>>, LlmError>> + Send + 'async_trait>>

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

Generate content as a stream of chunks, in the order the model produced them. LlmResponse::append folds them into the whole reply; usage, when a chunk carries it, is cumulative.