Struct SandboxTool 

pub struct SandboxTool { /* private fields */ }

Available on crate feature sandbox only.

A tool that executes code in an isolated sandbox.

SandboxTool wraps a SandboxBackend and implements adk_core::Tool, making sandbox execution available to LLM agents. The tool accepts language, code, optional stdin, and optional timeout_secs parameters.

Error Handling

Backend errors are never propagated as ToolError. Instead, they are converted to structured JSON with a "status" field ("timeout", "memory_exceeded", or "error"). This lets the agent reason about failures without triggering exception handling.

Example

use adk_sandbox::{SandboxTool, ProcessBackend};
use std::sync::Arc;

let backend = Arc::new(ProcessBackend::default());
let tool = SandboxTool::new(backend);
assert_eq!(tool.name(), "sandbox_exec");

Implementations

impl SandboxTool

pub fn new(backend: Arc<dyn SandboxBackend>) -> SandboxTool

Creates a new SandboxTool wrapping the given backend.

Trait Implementations

impl Tool for SandboxTool

fn name(&self) -> &str

fn description(&self) -> &str

fn required_scopes(&self) -> &[&str]

Returns the scopes required to execute this tool.

fn parameters_schema(&self) -> Option<Value>

fn execute<'life0, 'async_trait>( &'life0 self, _ctx: Arc<dyn ToolContext>, args: Value, ) -> Pin<Box<dyn Future<Output = Result<Value, AdkError>> + Send + 'async_trait>>

where 'life0: 'async_trait, SandboxTool: 'async_trait,

fn declaration(&self) -> Value

Returns the tool declaration that should be exposed to model providers.

fn enhanced_description(&self) -> String

Returns an enhanced description that may include additional notes. For long-running tools, this includes a warning not to call the tool again if it has already returned a pending status. Default implementation returns the base description.

fn is_long_running(&self) -> bool

Indicates whether the tool is a long-running operation. Long-running tools typically return a task ID immediately and complete the operation asynchronously.

fn is_builtin(&self) -> bool

Indicates whether this tool is a built-in server-side tool (e.g., google_search, url_context).

fn response_schema(&self) -> Option<Value>

fn is_read_only(&self) -> bool

Indicates whether this tool performs no side effects. Read-only tools may be executed concurrently in Auto mode.

fn is_concurrency_safe(&self) -> bool

Indicates whether this tool is safe for concurrent execution. Used by the Parallel strategy to validate dispatch safety.