Struct LlmAgentConfig 

pub struct LlmAgentConfig {
    pub system_prompt: String,
    pub tools: Vec<Arc<dyn Tool>>,
    pub lens: Lens,
    pub llm_config: LlmConfig,
    pub experience_extractor: Option<Arc<dyn ExperienceExtractor>>,
    pub refresh_every_n_tool_calls: Option<usize>,
}

Configuration for an LLM-powered agent.

Clone is supported: tools and extractors are Arc-wrapped for cheap sharing across workflow iterations (Loop) and concurrent tasks (Parallel).

Fields

system_prompt: String

System prompt that specializes this agent’s behavior.

tools: Vec<Arc<dyn Tool>>

Tools this agent can invoke during the Act phase.

lens: Lens

How this agent perceives the substrate.

llm_config: LlmConfig

Which LLM provider and model to use.

experience_extractor: Option<Arc<dyn ExperienceExtractor>>

Optional override for experience extraction logic. None uses the framework’s default extractor.

refresh_every_n_tool_calls: Option<usize>

How often to re-perceive the substrate during the Think→Act loop.

When set to Some(n), the agent re-runs the Perceive phase every n tool calls, picking up new experiences recorded by other agents in the same collective. This enables real-time “shared consciousness” in parallel workflows.

None disables mid-task refresh (perceive only once at start).

Trait Implementations

impl Clone for LlmAgentConfig

fn clone(&self) -> LlmAgentConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.