agents-core/ai.koog.agents.core.environment/SafeTool

SafeTool

data class SafeTool<TArgs : Tool.Args, TResult : ToolResult>(tool: Tool<TArgs, TResult>, environment: AIAgentEnvironment)(source)

A wrapper class designed to safely execute a tool within a given AI agent environment. It provides mechanisms for handling tool execution results and differentiating between success and failure cases.

Parameters

TArgs

The type of arguments accepted by the underlying tool. Must extend Tool.Args.

TResult

The type of result produced by the underlying tool. Must extend ToolResult.

Constructors

SafeTool

constructor(tool: Tool<TArgs, TResult>, environment: AIAgentEnvironment)

Types

Result

sealed interface Result<TResult : ToolResult>

Represents a sealed interface for results, which can either be a success or a failure.

Functions

execute

suspend fun execute(args: TArgs): SafeTool.Result<TResult>

Executes the tool with the provided arguments and returns the result.

executeRaw

suspend fun executeRaw(args: TArgs): String

Executes a raw tool call with the given arguments, returning the resulting content as a string.

executeUnsafe

suspend fun executeUnsafe(args: Tool.Args): SafeTool.Result<TResult>

Executes a tool with the provided arguments in an unsafe manner. This method does not enforce type safety for the arguments provided to the tool.