Struct AgentConfig 

pub struct AgentConfig {

    pub name: String,
    pub model: Option<String>,
    pub instruction: Option<String>,
    pub description: Option<String>,
    pub tools: Vec<ToolConfig>,
    pub sub_agents: Vec<AgentConfig>,
    pub temperature: Option<f32>,
    pub max_output_tokens: Option<u32>,
    pub thinking_budget: Option<u32>,
    pub output_key: Option<String>,
    pub output_schema: Option<Value>,
    pub max_llm_calls: Option<u32>,
    pub agent_type: String,
    pub max_iterations: Option<u32>,
    pub metadata: HashMap<String, Value>,
    pub voice: Option<String>,
    pub greeting: Option<String>,
    pub transcription: Option<bool>,
    pub a2a: Option<bool>,
    pub env: HashMap<String, String>,
}

Declarative agent configuration — loadable from YAML or TOML.

Example YAML

name: weather_agent
model: gemini-2.0-flash
instruction: "You are a helpful weather assistant."
description: "Provides weather information for cities."
tools:
  - name: get_weather
    description: "Get weather for a city"
  - builtin: google_search
sub_agents:
  - name: forecast_agent
    model: gemini-2.0-flash
    instruction: "Provide 5-day forecasts."
output_key: weather_result

Fields

name: String

Agent name (required).

model: Option<String>

Model identifier (e.g., “gemini-2.0-flash”, “gemini-2.5-pro”).

instruction: Option<String>

System instruction for the agent.

description: Option<String>

Human-readable description of what this agent does.

tools: Vec<ToolConfig>

Tool declarations.

sub_agents: Vec<AgentConfig>

Sub-agent configurations (for multi-agent hierarchies).

temperature: Option<f32>

Temperature for generation (0.0 - 2.0).

max_output_tokens: Option<u32>

Maximum output tokens.

thinking_budget: Option<u32>

Thinking budget (Google AI only).

output_key: Option<String>

State key to auto-save the agent’s final response into.

output_schema: Option<Value>

JSON Schema for structured output.

max_llm_calls: Option<u32>

Maximum number of LLM calls per invocation (safety limit).

agent_type: String

Agent type: “llm” (default), “sequential”, “parallel”, “loop”.

max_iterations: Option<u32>

For loop agents: maximum iterations.

metadata: HashMap<String, Value>

Custom metadata (passed through to state or callbacks).

voice: Option<String>

Voice configuration for live agents.

greeting: Option<String>

Greeting message (model speaks first on connect).

transcription: Option<bool>

Whether to enable transcription.

a2a: Option<bool>

Whether to enable A2A protocol endpoint.

env: HashMap<String, String>

Environment variables to set when loading this agent.

Implementations

impl AgentConfig

pub fn from_yaml_file(path: &Path) -> Result<Self, AgentConfigError>

Load agent config from a YAML file.

pub fn from_yaml(yaml: &str) -> Result<Self, AgentConfigError>

Parse agent config from a YAML string.

pub fn from_json(json: &str) -> Result<Self, AgentConfigError>

Parse agent config from a JSON string.

pub fn from_value(value: Value) -> Result<Self, AgentConfigError>

Parse agent config from a JSON value.

pub fn validate(&self) -> Result<(), AgentConfigError>

Validate the configuration.

pub fn builtin_tools(&self) -> Vec<&str>

Check if this is a built-in tool reference.

pub fn is_workflow(&self) -> bool

Check if this is a workflow agent (non-LLM).

Trait Implementations

impl Clone for AgentConfig

fn clone(&self) -> AgentConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for AgentConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for AgentConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for AgentConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.