Struct AgentRunOverrides 

pub struct AgentRunOverrides {
    pub security: Option<Arc<SecurityPolicy>>,
    pub memory: Option<Arc<dyn Memory>>,
    pub is_subagent: bool,
}

Optional per-call overrides for run.

SubAgent spawn paths use this to inject the validated child policy returned from SecurityPolicy::ensure_no_escalation_beyond (and, once v0.8.1 plumbs caller-supplied allowlist narrowing, the validated agent-scoped memory wrapper). Without this hook the run path rebuilds both surfaces from config, so the validator’s guarantees never reach the agent loop. None on either field preserves the from-config behavior — the same shape as a fresh interactive launch.

Fields

security: Option<Arc<SecurityPolicy>>

memory: Option<Arc<dyn Memory>>

is_subagent: bool

true when the run is a SubAgent invocation. SubAgents must not spawn further subagents (depth-1 cap). The agent loop reads this when constructing the spawn_subagent tool so the depth-cap refusal fires at the tool, not after a child run is already underway. Default false keeps top-level / cron-launched / CLI-launched agents at depth 0.

Trait Implementations

impl Default for AgentRunOverrides

fn default() -> AgentRunOverrides

Returns the “default value” for a type.