Struct Agent 

pub struct Agent<M>
where
    M: ModelAdapter,
{ /* private fields */ }

A configured agent ready to start a session.

Build one with Agent::builder, supplying at minimum a ModelAdapter. Optionally preload prior conversation state via AgentBuilder::transcript and the next user turn via AgentBuilder::input. Then call Agent::start with a SessionConfig to obtain a LoopDriver that drives the agentic loop.

If no input is preloaded, the first call to LoopDriver::next yields LoopInterrupt::AwaitingInput so the host can supply the first user message via InputRequest::submit. If input was preloaded, the first next() dispatches the model directly.

Example

use agentkit_core::{Item, ItemKind};
use agentkit_loop::{
    Agent, PromptCacheRequest, PromptCacheRetention, SessionConfig,
};
use agentkit_tools_core::ToolRegistry;

let agent = Agent::builder()
    .model(adapter)
    .add_tool_source(ToolRegistry::new())
    .transcript(vec![Item::text(ItemKind::System, "You are a helpful assistant.")])
    .input(vec![Item::text(ItemKind::User, "Hello!")])
    .build()?;

let mut driver = agent
    .start(SessionConfig::new("s1").with_cache(
        PromptCacheRequest::automatic().with_retention(PromptCacheRetention::Short),
    ))
    .await?;

// First next() drives the model since input was preloaded.
let _ = driver.next().await?;

Implementations

impl<M> Agent<M>

where M: ModelAdapter,

pub fn builder() -> AgentBuilder<M>

Create a new AgentBuilder for configuring this agent.

pub async fn start( &self, config: SessionConfig, ) -> Result<LoopDriver<M::Session>, LoopError>

Start a session, returning a LoopDriver preloaded with whatever transcript and input were configured on the builder. See AgentBuilder::transcript and AgentBuilder::input for what each one does and when to use them.

This calls ModelAdapter::start_session and emits an AgentEvent::RunStarted event to all registered observers.

&self so a single configured agent can mint multiple sessions over its lifetime — e.g. an outer agent that uses an inner sub-agent for transcript compaction.

Errors

Returns LoopError if the model adapter fails to create a session.