Trait CanUseToolCallback 

pub trait CanUseToolCallback: Send + Sync {
    // Required method
    fn call<'a>(
        &'a self,
        ctx: ToolPermissionContext,
    ) -> Pin<Box<dyn Future<Output = PermissionDecision> + Send + 'a>>;
}

Trait for permission callbacks.

Implementations are typically async closures or plain functions. The SDK wraps them in a boxed trait object inside Options.

Panics and error handling

Callbacks are expected to never panic. If a callback does panic, the tokio task running the current next_event call is aborted; the next call to next_event returns an Error::Io with a broken-pipe or similar message. Authors should return PermissionDecision::deny to signal rejection rather than panicking.

Callbacks cannot signal I/O or other errors. If your callback performs fallible work (e.g., consulting a policy server), handle the failure internally and translate to allow or deny(reason) - the SDK does not surface callback errors to the claude binary separately from a deny response.

Required Methods

fn call<'a>( &'a self, ctx: ToolPermissionContext, ) -> Pin<Box<dyn Future<Output = PermissionDecision> + Send + 'a>>

Called by the SDK when the claude binary requests permission.

Dyn Compatibility

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors

impl<F, Fut> CanUseToolCallback for F

where F: Fn(ToolPermissionContext) -> Fut + Send + Sync, Fut: Future<Output = PermissionDecision> + Send + 'static,