how to stop the execution of an agent inside a dynamic tool?

[valdo99]

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

typescript

import { DynamicStructuredTool, DynamicTool } from '@langchain/core/tools';

export function addItemToCart() {
    const add_to_cart = new DynamicStructuredTool({
        name: 'add_to_cart',
        description:
            'Adds a product to the cart. This function processes a product ID, variant ID and a quantity to add the product to the cart. the product ID and variant ID must be taken from the option available in the "context <ctx>". ',
        func: async ({ variant_id, quantity }) => {
            try {
               //... retrieve cart

                if (!cart) {
                   // HERE I NEED TO STOP THE AGENT EXECUTOR
                }
              return cart.total
            } catch (error) {
                console.error('Error adding to cart', error);
                return 'Error adding the product to the cart';
            }
        },
        schema: z
            .object({
                variant_id: z
                    .string()
                    .describe(
                        'A unique identifier for a specific product variant, used to differentiate versions of a product based on attributes like size or color. this field must be taken from the context <ctx>.',
                    ),
                quantity: z
                    .number()
                    .optional()
                    .default(1)
                    .describe(
                        'Specifies the quantity of the product variant to be added to the cart. Defaults to 1 if not specified; field is optional.',
                    ),
            })
            .describe(
                'Defines the json schema for the product information required to add a product to the cart. This schema ensures that the correct data format and types are adhered to during the addition process.',
            ),
    });

    return add_to_cart;
}

Description

im using langchain js to create an agent for shopping.

but in some cases i need to stop the agent executor because i have custom business logics that doesnt need anymore the generation...

System Info

langchain@0.2.4 | MIT | deps: 16 | versions: 276
Typescript bindings for langchain
https://github.com/langchain-ai/langchainjs/tree/main/langchain/

keywords: llm, ai, gpt3, chain, prompt, prompt engineering, chatgpt, machine learning, ml, openai, embeddings, vectorstores

dist
.tarball: https://registry.npmjs.org/langchain/-/langchain-0.2.4.tgz
.shasum: a85295e2510425cc5f4aca32bc9adeda88f6918d
.integrity: sha512-zBsBuNREn/3IlWvIQqhQ2iqf6JJhyjjsB1Db/keDkcgThPI3EcblC1pqAXU2BIKHmpNUkHBR2bAUok5+xtgOcw==
.unpackedSize: 4.0 MB

dependencies:
@langchain/core: ~0.2.0 js-tiktoken: ^1.0.12 langsmith: ~0.1.30 uuid: ^9.0.0
@langchain/openai: ~0.1.0 js-yaml: ^4.1.0 ml-distance: ^4.0.0 yaml: ^2.2.1
@langchain/textsplitters: ~0.0.0 jsonpointer: ^5.0.1 openapi-types: ^12.1.3 zod-to-json-schema: ^3.22.3
binary-extensions: ^2.2.0 langchainhub: ~0.0.8 p-retry: 4 zod: ^3.22.4

maintainers:

• nfcampos nuno@boringbits.io
• jacoblee93 jacoblee93@gmail.com
• andrewnguonly andrewnguonly@gmail.com
• davidduong david@duong.cz
• hwchase17 hw.chase.17@gmail.com
• basproul braceasproul@gmail.com

dist-tags:
latest: 0.2.4 next: 0.2.3-rc.0

published 16 hours ago by jacoblee93 jacoblee93@gmail.com

[dosubot[bot]]

To stop the execution of an agent inside a dynamic tool in LangChain JS, you can use the AbortController to cancel the request. Here is an example of how to do it:

import { initializeAgentExecutorWithOptions } from "langchain/agents";
import { OpenAI } from "@langchain/openai";
import { Calculator } from "@langchain/community/tools/calculator";
import { SerpAPI } from "@langchain/community/tools/serpapi";

const model = new OpenAI({ temperature: 0 });
const tools = [
  new SerpAPI(process.env.SERPAPI_API_KEY, {
    location: "Austin,Texas,United States",
    hl: "en",
    gl: "us",
  }),
  new Calculator(),
];
const executor = await initializeAgentExecutorWithOptions(tools, model, {
  agentType: "zero-shot-react-description",
});
const controller = new AbortController();

// Call `controller.abort()` somewhere to cancel the request.
setTimeout(() => {
  controller.abort();
}, 2000);

try {
  const input = `Who is Olivia Wilde's boyfriend? What is his current age raised to the 0.23 power?`;
  const result = await executor.invoke({ input, signal: controller.signal });
} catch (e) {
  console.log(e);
  /*
  Error: Cancel: canceled
      at file:///Users/nuno/dev/langchainjs/langchain/dist/util/async_caller.js:60:23
      at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
      at RetryOperation._fn (/Users/nuno/dev/langchainjs/node_modules/p-retry/index.js:50:12) {
    attemptNumber: 1,
    retriesLeft: 6
  }
  */
}

In this example, the controller.abort() method is called after 2 seconds, which will cancel the ongoing request. Additionally, the AgentExecutor class in LangChain JS has built-in support for stopping execution based on conditions such as reaching the final output or exceeding the maximum number of iterations. The _stop method in the AgentExecutor class can be used to handle such scenarios.

Note that this will only cancel the outgoing request if the underlying provider exposes that option. LangChain will cancel the underlying request if possible; otherwise, it will cancel the processing of the response ^[1][2][3].

To continue talking to Dosu, mention @dosu.

[buddypia]

@valdo99 What happens if I use return_direct?
Please refer to this.