Struct OpenAiToolAgent 

pub struct OpenAiToolAgent { /* private fields */ }

Implementations

impl OpenAiToolAgent

pub fn create_prompt(prefix: &str) -> Result<MessageFormatterStruct, AgentError>

Trait Implementations

impl Agent for OpenAiToolAgent

fn plan_stream<'a>( &'a self, intermediate_steps: &'a [(AgentAction, String)], inputs: PromptArgs, ) -> PlanEventStream<'a>

Streams the model’s text as it arrives (with native tool calling that is the answer, or a short preamble like “Let me check” before tool calls).

fn plan<'life0, 'life1, 'async_trait>( &'life0 self, intermediate_steps: &'life1 [(AgentAction, String)], inputs: PromptArgs, ) -> Pin<Box<dyn Future<Output = Result<AgentEvent, AgentError>> + Send + 'async_trait>>

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

fn plan_with_usage<'life0, 'life1, 'async_trait>( &'life0 self, intermediate_steps: &'life1 [(AgentAction, String)], inputs: PromptArgs, ) -> Pin<Box<dyn Future<Output = Result<(AgentEvent, Option<TokenUsage>), AgentError>> + Send + 'async_trait>>

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

Like plan, and also returns the tokens the model call(s) used, so AgentExecutor can report a run’s total.

fn get_tools(&self) -> Vec<Arc<dyn Tool>>