Enum LoopInterrupt 

pub enum LoopInterrupt {
    ApprovalRequest(PendingApproval),
    AuthRequest(PendingAuth),
    AwaitingInput(InputRequest),
    AfterToolResult(ToolRoundInfo),
}

An interrupt that pauses the agent loop until the host resolves it.

The loop returns an interrupt inside LoopStep::Interrupt whenever it cannot proceed autonomously. Each variant carries a handle with resolution methods so callers can resolve the interrupt directly.

Example

use agentkit_loop::{LoopInterrupt, LoopStep};

match driver.next().await? {
    LoopStep::Interrupt(LoopInterrupt::ApprovalRequest(pending)) => {
        println!("Tool {} needs approval: {}", pending.request.request_kind, pending.request.summary);
        pending.approve(driver)?;
    }
    LoopStep::Interrupt(LoopInterrupt::AuthRequest(pending)) => {
        println!("Auth required from provider: {}", pending.request.provider);
        pending.cancel(driver)?;
    }
    LoopStep::Interrupt(LoopInterrupt::AwaitingInput(pending)) => {
        println!("Waiting for input: {}", pending.reason);
        // ... call pending.submit(driver, items)
    }
    LoopStep::Interrupt(LoopInterrupt::AfterToolResult(info)) => {
        // Cooperative yield between tool rounds.  Optionally call
        // driver.submit_input(...) to interject a user message, then
        // call driver.next() to resume the turn.
        let _ = info;
    }
    LoopStep::Finished(result) => {
        println!("Turn finished: {:?}", result.finish_reason);
    }
}

Variants

ApprovalRequest(PendingApproval)

A tool call requires explicit approval before it can execute.

AuthRequest(PendingAuth)

A tool call requires authentication credentials.

AwaitingInput(InputRequest)

The driver has no pending input and needs the host to supply some.

AfterToolResult(ToolRoundInfo)

A tool round finished: all tool calls from the previous assistant message now have results in the transcript, and the driver is about to invoke the model again. The host may call LoopDriver::submit_input to interject user messages before the next model turn, then call LoopDriver::next to resume.

This is a non-blocking interrupt: callers that do not care about mid-turn interjection can treat it as a no-op (_ => continue) and the next next() call resumes the turn.

Implementations

impl LoopInterrupt

pub fn is_blocking(&self) -> bool

Returns true if the interrupt must be explicitly resolved before the loop can make progress. Approvals and auth requests are blocking; AwaitingInput and AfterToolResult are cooperative and can be ignored by calling LoopDriver::next again.

Trait Implementations

impl Clone for LoopInterrupt

fn clone(&self) -> LoopInterrupt

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LoopInterrupt

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for LoopInterrupt

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for LoopInterrupt

fn eq(&self, other: &LoopInterrupt) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for LoopInterrupt

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl Eq for LoopInterrupt

impl StructuralPartialEq for LoopInterrupt