problem with use returnDirect: true on react agent

[miniboxhaha]

React and CodeAct are often used together.
CodeAct makes several attempts until it finally produces executable code.
when use returnDirect on React, it use the first try as result.
The code sandbox's return value allows me to verify whether the code executed as intended.
so, is there possible to control returnDirect value from tool definition

import { createReactAgent } from "@langchain/langgraph/prebuilt";
import { initChatModel } from "langchain/chat_models/universal";
import { tool } from "@langchain/core/tools";
import { z } from "zod";

const add = tool(
  async (input: { code }) => {
    cosnt r = callsandbox(code);
    if(r.status.ok){
        return {result:r.result, returnDirect:true}
    }
    return {result:r.error, returnDirect:false}
  },
  {
    name: "add",
    schema: z.object({
      a: z.number().describe("First operand"),
      b: z.number().describe("Second operand"),
    }),
    description: "Add two numbers.",
    returnDirect: var,
  }
);

const llm = await initChatModel("anthropic:claude-3-7-sonnet-latest");
const agent = createReactAgent({
  llm,
  tools: [add],
});

const response = await agent.invoke({
  messages: [{ role: "user", content: "what's 3 + 5?" }],
});