FunctionToolResult

type FunctionToolResult<Context, TParameters, Result> =
  | {
  agentRunResult?: RunResult<Context, Agent<Context, any>>;
  interruptions?: RunToolApprovalItem[];
  output: string | unknown;
  runItem: RunToolCallOutputItem;
  tool: FunctionTool<Context, TParameters, Result>;
  type: "function_output";
}
  | {
  runItem: RunToolApprovalItem;
  tool: FunctionTool<Context, TParameters, Result>;
  type: "function_approval";
}
  | {
  runItem: RunToolApprovalItem;
  tool: HostedMCPTool<Context>;
  type: "hosted_mcp_tool_approval";
};

The result of invoking a function tool. Either the actual output of the execution or a tool approval request.

These get passed for example to the toolUseBehavior option of the Agent constructor.

Type Parameters

Type Parameter	Default type
Context	UnknownContext
TParameters extends ToolInputParameters	any
Result	any

Type declaration

{
  agentRunResult?: RunResult<Context, Agent<Context, any>>;
  interruptions?: RunToolApprovalItem[];
  output: string | unknown;
  runItem: RunToolCallOutputItem;
  tool: FunctionTool<Context, TParameters, Result>;
  type: "function_output";
}

agentRunResult?

optional agentRunResult: RunResult<Context, Agent<Context, any>>;

The result returned when the tool execution runs another agent. Populated when the invocation originated from Agent.asTool and the nested agent completed a run.

interruptions?

optional interruptions: RunToolApprovalItem[];

Any interruptions collected while the nested agent executed. These are surfaced to allow callers to pause and resume workflows that require approvals.

output

output: string | unknown;

The output of the tool call. This can be a string or a stringifable item.

runItem

runItem: RunToolCallOutputItem;

The run item representing the tool call output.

tool

tool: FunctionTool<Context, TParameters, Result>;

The tool that was called.

type

type: "function_output";

{
  runItem: RunToolApprovalItem;
  tool: FunctionTool<Context, TParameters, Result>;
  type: "function_approval";
}

runItem

runItem: RunToolApprovalItem;

The item representing the tool call that is requiring approval.

tool

tool: FunctionTool<Context, TParameters, Result>;

The tool that is requiring to be approved.

type

type: "function_approval";

Indicates that the tool requires approval before it can be called.

{
  runItem: RunToolApprovalItem;
  tool: HostedMCPTool<Context>;
  type: "hosted_mcp_tool_approval";
}

runItem

runItem: RunToolApprovalItem;

The item representing the tool call that is requiring approval.

tool

tool: HostedMCPTool<Context>;

The tool that is requiring to be approved.

type

type: "hosted_mcp_tool_approval";

Indicates that the tool requires approval before it can be called.