Skip to main content

AgentRuntime

Trait AgentRuntime 

Source
pub trait AgentRuntime: Send + Sync {
    // Provided methods
    fn phase(&self) -> &str { ... }
    fn threat_score(&self, _path: &str) -> f32 { ... }
    fn sender_trust(&self, _email: &str) -> SenderTrust { ... }
    fn skill_name(&self) -> &str { ... }
    fn inbox_count(&self) -> usize { ... }
    fn has_otp(&self) -> bool { ... }
    fn has_threat(&self) -> bool { ... }
    fn intent(&self) -> &str { ... }
    fn context_summary(&self) -> String { ... }
}
Expand description

Runtime context available to the agent during execution. Implement this to connect your pipeline, classifiers, and state machines.

Provided Methods§

Source

fn phase(&self) -> &str

Current workflow phase (e.g. “Reading”, “Acting”, “Cleanup”).

Source

fn threat_score(&self, _path: &str) -> f32

Threat score for a specific file (0.0 = safe, 1.0 = attack).

Source

fn sender_trust(&self, _email: &str) -> SenderTrust

Sender trust level for an email address.

Source

fn skill_name(&self) -> &str

Currently selected skill name (from classifier).

Source

fn inbox_count(&self) -> usize

Number of inbox files in current trial.

Source

fn has_otp(&self) -> bool

Whether any inbox file contains OTP/credential content.

Source

fn has_threat(&self) -> bool

Whether any inbox file has injection/threat signals.

Source

fn intent(&self) -> &str

Classified intent (e.g. “intent_inbox”, “intent_delete”).

Source

fn context_summary(&self) -> String

Structured summary for think tool context injection.

Implementors§