error type in create a ReAct agent from scratch

[ixn3rd3mxn]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

import * as dotenv from "dotenv";
dotenv.config();

import { ChatOpenAI } from "@langchain/openai";
import { tool } from "@langchain/core/tools";
import { z } from "zod";

const model = new ChatOpenAI({
  modelName: "gpt-4o-mini",
  configuration: {
    baseURL: process.env.OPENAI_API_BASE_URL,
    apiKey: process.env.OPENAI_API_KEY,
  }
});

const getWeather = tool(async ({ location }) => {
  const lowercaseLocation = location.toLowerCase();
  if (lowercaseLocation.includes("sf") || lowercaseLocation.includes("san francisco")) {
    return "It's sunny!";
  } else if (lowercaseLocation.includes("boston")) {
    return "It's rainy!";
  } else {
    return `I am not sure what the weather is in ${location}`;
  }
}, {
  name: "getWeather",
  schema: z.object({
    location: z.string().describe("location to get the weather for"),
  }),
  description: "Call to get the weather from a specific location."
});

const tools = [getWeather];

import {
  type BaseMessageLike,
  AIMessage,
  ToolMessage,
} from "@langchain/core/messages";
import { type ToolCall } from "@langchain/core/messages/tool";
import { task } from "@langchain/langgraph";

const toolsByName = Object.fromEntries(tools.map((tool) => [tool.name, tool]));

const callModel = task("callModel", async (messages: BaseMessageLike[]) => {
  const response = await model.bindTools(tools).invoke(messages);
  return response;
});

const callTool = task(
  "callTool",
  async (toolCall: ToolCall): Promise<AIMessage> => {
    const tool = toolsByName[toolCall.name];
    const observation = await tool.invoke(toolCall.args);
    return new ToolMessage({ content: observation, tool_call_id: toolCall.id });
    // Can also pass toolCall directly into the tool to return a ToolMessage
    // return tool.invoke(toolCall);
  });

  import { entrypoint, addMessages } from "@langchain/langgraph";

  const agent = entrypoint(
    "agent",
    async (messages: BaseMessageLike[]) => {
      let currentMessages = messages;
      let llmResponse = await callModel(currentMessages);
      while (true) {
        if (!llmResponse.tool_calls?.length) {
          break;
        }
  
        // Execute tools
        const toolResults = await Promise.all(
          llmResponse.tool_calls.map((toolCall) => {
            return callTool(toolCall);
          })
        );
  
        // Append to message list
        currentMessages = addMessages(currentMessages, [llmResponse, ...toolResults]);
  
        // Call model again
        llmResponse = await callModel(currentMessages);
      }
  
      return llmResponse;
    }
  );

  import { BaseMessage, isAIMessage } from "@langchain/core/messages";

  const prettyPrintMessage = (message: BaseMessage) => {
    console.log("=".repeat(30), `${message.getType()} message`, "=".repeat(30));
    console.log(message.content);
    if (isAIMessage(message) && message.tool_calls?.length) {
      console.log(JSON.stringify(message.tool_calls, null, 2));
    }
  }
  
  // Usage example
  const userMessage = { role: "user", content: "hello test test" };
  console.log(userMessage);
  
  const stream = await agent.stream([userMessage]);
  
  for await (const step of stream) {
    for (const [taskName, update] of Object.entries(step)) {
      const message = update as BaseMessage;
      // Only print task updates
      if (taskName === "agent") continue;
      console.log(`\n${taskName}:`);
      prettyPrintMessage(message);
    }
  }

Description

my code i follow in this link How to create a ReAct agent from scratch

my code is have error 3

[{
	"resource": "/c:/Users/Administrator/langgraph-agent/main_custom.ts",
	"owner": "typescript",
	"code": "2345",
	"severity": 8,
	"message": "Argument of type 'Record<string, any>' is not assignable to parameter of type 'StructuredToolCallInput<ZodObject<{ location: ZodString; }, \"strip\", ZodTypeAny, { location: string; }, { location: string; }>, { location: string; }>'.\n  Type 'Record<string, any>' is missing the following properties from type 'ToolCall': name, args",
	"source": "ts",
	"startLineNumber": 54,
	"startColumn": 43,
	"endLineNumber": 54,
	"endColumn": 56
},{
	"resource": "/c:/Users/Administrator/langgraph-agent/main_custom.ts",
	"owner": "typescript",
	"code": "2322",
	"severity": 8,
	"message": "Type 'string | ToolMessage' is not assignable to type 'string | (MessageContentComplex | DataContentBlock)[]'.\n  Type 'ToolMessage' is not assignable to type 'string | (MessageContentComplex | DataContentBlock)[]'.",
	"source": "ts",
	"startLineNumber": 55,
	"startColumn": 30,
	"endLineNumber": 55,
	"endColumn": 37,
	"relatedInformation": [
		{
			"startLineNumber": 4,
			"startColumn": 5,
			"endLineNumber": 4,
			"endColumn": 12,
			"message": "The expected type comes from property 'content' which is declared here on type 'ToolMessageFieldsWithToolCallId'",
			"resource": "/c:/Users/Administrator/langgraph-agent/node_modules/@langchain/core/dist/messages/tool.d.ts"
		}
	]
},{
	"resource": "/c:/Users/Administrator/langgraph-agent/main_custom.ts",
	"owner": "typescript",
	"code": "2322",
	"severity": 8,
	"message": "Type 'string | undefined' is not assignable to type 'string'.\n  Type 'undefined' is not assignable to type 'string'.",
	"source": "ts",
	"startLineNumber": 55,
	"startColumn": 52,
	"endLineNumber": 55,
	"endColumn": 64,
	"relatedInformation": [
		{
			"startLineNumber": 15,
			"startColumn": 5,
			"endLineNumber": 15,
			"endColumn": 17,
			"message": "The expected type comes from property 'tool_call_id' which is declared here on type 'ToolMessageFieldsWithToolCallId'",
			"resource": "/c:/Users/Administrator/langgraph-agent/node_modules/@langchain/core/dist/messages/tool.d.ts"
		}
	]
}]

but it work when i try run this

C:\Users\Administrator\langgraph-agent>npx tsx main_custom.ts
{ role: 'user', content: 'weather boston' }

callModel:
============================== ai message ==============================

[
  {
    "name": "getWeather",
    "args": {
      "location": "Boston"
    },
    "type": "tool_call",
    "id": "call_NH8HerfgYo8NVmNokKRU4Ck0"
  }
]

callTool:
============================== tool message ==============================
It's rainy!

callModel:
============================== ai message ==============================
The weather in Boston is rainy.

In conclusion, what are the effects of this error? Is it necessary to fix it? Because the code runs normally. If it needs to be fixed, how should it be fixed?

System Info

[email protected]
platform windows 11
Node.js v22.15.0
npm: '10.9.2'

[dosubot[bot]]

Hello @ixn3rd3mxn! I'm Dosu, a bot here to help you solve bugs, answer questions, and assist you in becoming a contributor. While you're waiting for a human maintainer, feel free to share any issues you're facing!

The errors you're encountering are TypeScript type errors, which are not preventing your code from running but indicate potential type mismatches that could lead to runtime errors. Here's a breakdown of the errors and their potential effects:

1. Error 2345:

   • Message: "Argument of type 'Record<string, any>' is not assignable to parameter of type 'StructuredToolCallInput'."
   • Cause: This error suggests that the toolCall object being passed to the callTool function does not match the expected type StructuredToolCallInput. This could lead to runtime errors if the toolCall object does not have the expected structure or properties.
   • Effect: If the toolCall object does not have the expected properties, it could cause the callTool function to behave unexpectedly or fail.

2. Error 2:

   • Message: "Type 'string | ToolMessage' is not assignable to type 'string | (MessageContentComplex | DataContentComplex)'."
   • Cause: This error indicates a type mismatch. The callTool function is expected to return an AIMessage, but it seems to be returning a ToolMessage or a string. The AIMessage type might not be compatible with the ToolMessage type, or the ToolMessage might not be correctly structured.
   • Effect: This could lead to issues when the callTool function is used, as the returned value might not be handled correctly by the rest of the code.

3. Error 3:

   • Message: "Type 'string | undefined' is not assignable to type 'string'."
   • Cause: This error suggests that the tool_call_id property of the ToolMessage is expected to be a string, but it might be undefined in some cases.
   • Effect: If tool_call_id is undefined, it could lead to issues when the ToolMessage is processed, as this property is used to track and reference the specific tool call associated with the message ^[1].

Conclusion: While your code seems to run without errors, these TypeScript errors indicate potential issues that could cause runtime errors or unexpected behavior. It is recommended to address these type mismatches to ensure the robustness and reliability of your code. You may need to review the structure of the toolCall object and ensure that the tool_call_id is always defined when creating a ToolMessage.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}