Struct ToolRpc 

pub struct ToolRpc { /* private fields */ }

Implementations

impl ToolRpc

pub fn new(tool_name: &str) -> Self

Binds an ordinary tool client to its configured tool name.

impl ToolRpc

pub fn create(tool_name: &str) -> Result<ToolRpc, ToolRpcError>

Creates a tool RPC resource without trapping on invalid caller input.

impl ToolRpc

pub async fn invoke_and_await( &self, command_path: Vec<String>, input: TypedSchemaValue, stdin: Option<ToolStdin>, stdout: Option<ToolOutput>, stderr: Option<ToolOutput>, ) -> Result<InvocationResult, ToolRpcError>

Waits for the structured terminal. Callers that supplied stdout or stderr must drive this wait and every already-created reader concurrently. Callers that manually created an open stdin must likewise drive its writer concurrently; see create-stdin.

impl ToolRpc

pub fn invoke( &self, command_path: &[String], input: TypedSchemaValue, stdin: Option<ToolStdin>, ) -> Result<(), ToolRpcError>

Durably admits fire-and-forget work. Declared output is discarded by the host so an absent caller reader cannot apply backpressure.

impl ToolRpc

pub fn async_invoke_and_await( &self, command_path: &[String], input: TypedSchemaValue, stdin: Option<ToolStdin>, stdout: Option<ToolOutput>, stderr: Option<ToolOutput>, ) -> FutureInvokeResult

Durably admits work and returns an independently owned structured result observer. Each requested output was created by the caller and is observed independently from this future.

Trait Implementations

impl Debug for ToolRpc

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

Formats the value using the given formatter.