Unable to dynamically register tools according to the documentation.

[kylectx]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangGraph.js documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangGraph.js rather than my code.
• The bug is not resolved by updating to the latest stable version of LangGraph (or the specific integration package).

Example Code

Here is the test code

import { createAgent, createMiddleware, tool } from "langchain";
import * as z from "zod";
import {ChatOpenAI} from "@langchain/openai";

const model = new ChatOpenAI({
    model: "my-model",
    apiKey: "my-api-key",
    streaming: true,
});

const getWeather = tool(
    ({ location }) => `Weather in ${location}: Sunny, 72°F`,
    {
        name: "get_weather",
        description: "Get weather information for a location",
        schema: z.object({
            location: z.string().describe("The location to get weather for"),
        }),
    }
);

// A tool that will be added dynamically at runtime
const calculateTip = tool(
    ({ billAmount, tipPercentage = 20 }) => {
        const tip = billAmount * (tipPercentage / 100);
        return `Tip: $${tip.toFixed(2)}, Total: $${(billAmount + tip).toFixed(2)}`;
    },
    {
        name: "calculate_tip",
        description: "Calculate the tip amount for a bill",
        schema: z.object({
            billAmount: z.number().describe("The bill amount"),
            tipPercentage: z.number().default(20).describe("Tip percentage"),
        }),
    }
);

const dynamicToolMiddleware = createMiddleware({
    name: "DynamicToolMiddleware",
    wrapModelCall: (request, handler) => {
        // Add dynamic tool to the request
        // This could be loaded from an MCP server, database, etc.
        return handler({
            ...request,
            tools: [...request.tools, calculateTip],
        });
    },
    wrapToolCall: (request, handler) => {
        // Handle execution of the dynamic tool
        if (request.toolCall.name === "calculate_tip") {
            return handler({ ...request, tool: calculateTip });
        }
        return handler(request);
    },
});

const agent = createAgent({
    model: model,
    tools: [getWeather], // Only static tools registered here
    middleware: [dynamicToolMiddleware],
});

// The agent can now use both getWeather AND calculateTip
const result = agent.invoke({
    messages: [{ role: "user", content: "Calculate a 20% tip on $85" }],
});

Error Message and Stack Trace (if applicable)

Uncaught Error: You have added a new tool in "wrapModelCall" hook of middleware "DynamicToolMiddleware": calculate_tip. This is not supported.

Description

I've just started learning langchain + langgraph. I'm currently using the TypeScript versions langchain 1.2.15 and langgraph 1.1.2. I wanted to follow this document to implement dynamic registration of tools: https://docs.langchain.com/oss/javascript/langchain/agents#runtime-tool-registration, but I get error.

I found that in langchain's source file AgentNode.js there is indeed logic that checks for newTools and invalidTools, which differs from what the documentation describes.

System Info

Node v20.18.0
packages
├── @langchain/core@1.1.17
├── @langchain/langgraph@1.1.2
├── @langchain/mcp-adapters@1.1.2
├── @langchain/openai@1.2.3

[pavel-voronin]

I reproduced it

What’s happening is simple:

• The docs show that you can add tools at runtime in middleware (wrapModelCall + wrapToolCall):
  Runtime tool registration
• In the current JS runtime, adding a new tool in wrapModelCall is blocked and throws exactly this error.

So right now, behavior is:

• wrapModelCall can be used to filter/select already registered tools:
  Filtering pre-registered tools
• But adding a brand-new tool there is rejected by runtime validation.

In short: this looks like a docs/runtime mismatch