Struct TaskOutputTool 

pub struct TaskOutputTool { /* private fields */ }

Implementations

impl TaskOutputTool

pub fn new(m: Arc<TaskManager>) -> Self

Trait Implementations

impl Tool for TaskOutputTool

fn definition(&self) -> ToolDefinition

Describe the tool’s name, prose description, and JSON-schema input.

fn is_concurrency_safe(&self, _: &Value) -> bool

Whether the tool is safe to run concurrently with other invocations.

fn invoke<'life0, 'async_trait>( &'life0 self, args: Value, __arg2: ToolContext, ) -> Pin<Box<dyn Future<Output = Result<ToolOutput, AgentError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Execute the tool. Errors are surfaced as is_error: true tool results rather than aborting the turn, so the model can try to recover.

fn prompt_text(&self) -> Option<&'static str>

Optional detailed usage instructions injected into the system prompt. Return None if the tool has no extra behavioral guidance.

fn validate<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, _arguments: &'life1 Value, _context: &'life2 ToolContext, ) -> Pin<Box<dyn Future<Output = Result<(), AgentError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Validate raw arguments before the permission check runs.

fn check_permission<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, _arguments: &'life1 Value, _context: &'life2 ToolContext, ) -> Pin<Box<dyn Future<Output = Result<PermissionDecision, AgentError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Decide whether the call is allowed, denied, or needs human approval.

fn interrupt_behavior(&self) -> InterruptBehavior

How the tool wants to be interrupted.

fn invocation_detail(&self, arguments: &Value) -> String

Return a human-readable summary of the invocation from its arguments.