Seems renderTextDescription lacks the description of tool arguments? which makes my reAct agent always fail

[danielwpz]

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

const tools = [tool(toolFn, toolSchema)];

  const llm = new SomeLLM();

  const prompt = await pull<PromptTemplate>("hwchase17/react");
  
  const agent = await createReactAgent({
    llm,
    tools,
    prompt,
  });
  
  const agentExecutor = new AgentExecutor({
    agent,
    tools,
  });
  
  const result = await agentExecutor.invoke({
    input: "use the tool to search for 'langchain'",
  });

  console.log(result);

Description

• I expect the react agent to automatically handles tool calling for me even if the llm doesn't support native tool calls
• However, since renderTextDescription doesn't contain descriptions of how to construct tool call arguments, the llm will return in random format, which cannot be correctly parsed later, and leads to an error ToolInputParsingException [Error]: Received tool input did not match expected schema
• I search for the same function in langchain-py, it will try to include the function signature in the tool description (https://github.com/langchain-ai/langchain/blob/master/libs/core/langchain_core/tools/render.py#L31)
• I'm wondering if langchian-js should follow the same logic? Or I need to manually do some extra work to tell llm how to respond with tool call?

System Info

• "@langchain/core": "^0.3.42",
• tsx v4.19.3
• node v20.12.2