Struct Agent 

pub struct Agent {
    pub name: String,
    pub llm: Arc<dyn LLM>,
    pub tools: HashMap<String, Arc<dyn Tool>>,
    pub system_prompt: Option<String>,
    pub memory: Vec<String>,
    pub max_iterations: usize,
}

High-level agent that holds an LLM and a set of tools, plus simple agent state.

Fields

name: String

A short, human-friendly name for the agent instance.

llm: Arc<dyn LLM>

The LLM implementation used to generate responses/thoughts.

tools: HashMap<String, Arc<dyn Tool>>

Registered tools the agent may call by name.

system_prompt: Option<String>

Optional system prompt / instructions provided to the LLM describing the agent’s role and available behaviors.

memory: Vec<String>

Simple short-term memory / conversation context kept by the agent. We store Message objects elsewhere in the crate; for a minimal implementation we keep user-visible strings here.

max_iterations: usize

Maximum iterations when running a looped decision process.

Implementations

impl Agent

pub fn new( name: impl Into<String>, llm: Arc<dyn LLM>, max_iterations: Option<usize>, ) -> Self

Create a new Agent with the provided name and LLM. Tools start empty.

pub fn register_tool( &mut self, name: Option<&str>, tool: Arc<dyn Tool>, ) -> &mut Self

Register a tool under the given name. Replaces any existing tool with the same name. Returns &mut Self for chaining.

pub fn change_max_iterations(&mut self, max_iterations: usize)

Change the maximum iterations for the agent’s decision process.

pub fn get_tool(&self, name: &str) -> Option<Arc<dyn Tool>>

Look up a tool by name.

pub fn set_system_prompt(&mut self, prompt: impl Into<String>)

Set or replace the agent’s system prompt.

pub fn generate_system_prompt(&self) -> Vec<Message>

pub fn generate_tools_prompt(&self) -> Vec<Message>

Trait Implementations

impl AgentRunner for Agent

fn call_llm<'life0, 'life1, 'async_trait>( &'life0 self, prompt: &'life1 str, ) -> Pin<Box<dyn Future<Output = AgentExecuteResult> + Send + 'async_trait>>

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

Call the LLM with a prompt and return the generation result.