Skip to main content

anthropic_tools

danger

This API is deprecated as Anthropic now officially supports tools. Click here to read the documentation.

Anthropic Tools

LangChain offers an experimental wrapper around Anthropic that gives it the same API as OpenAI Functions.

Setup​

To start, install the @langchain/anthropic integration package.

tip

See this section for general instructions on installing integration packages.

npm install @langchain/anthropic

Initialize model​

You can initialize this wrapper the same way you'd initialize a standard ChatAnthropic instance:

tip

We're unifying model params across all packages. We now suggest using model instead of modelName, and apiKey for API keys.

import { ChatAnthropicTools } from "@langchain/anthropic/experimental";

const model = new ChatAnthropicTools({
  temperature: 0.1,
  model: "claude-3-sonnet-20240229",
  apiKey: "YOUR-API-KEY", // In Node.js defaults to process.env.ANTHROPIC_API_KEY
});

Passing in tools​

You can now pass in tools the same way as OpenAI:

import { ChatAnthropicTools } from "@langchain/anthropic/experimental";
import { HumanMessage } from "@langchain/core/messages";

const model = new ChatAnthropicTools({
  temperature: 0.1,
  model: "claude-3-sonnet-20240229",
}).bind({
  tools: [
    {
      type: "function",
      function: {
        name: "get_current_weather",
        description: "Get the current weather in a given location",
        parameters: {
          type: "object",
          properties: {
            location: {
              type: "string",
              description: "The city and state, e.g. San Francisco, CA",
            },
            unit: { type: "string", enum: ["celsius", "fahrenheit"] },
          },
          required: ["location"],
        },
      },
    },
  ],
  // You can set the `function_call` arg to force the model to use a function
  tool_choice: {
    type: "function",
    function: {
      name: "get_current_weather",
    },
  },
});

const response = await model.invoke([
  new HumanMessage({
    content: "What's the weather in Boston?",
  }),
]);

console.log(response);

/*
  AIMessage {
    lc_serializable: true,
    lc_kwargs: { content: '', additional_kwargs: { tool_calls: [Array] } },
    lc_namespace: [ 'langchain_core', 'messages' ],
    content: '',
    name: undefined,
    additional_kwargs: { tool_calls: [ [Object] ] }
  }
*/

console.log(response.additional_kwargs.tool_calls);

/*
  [
    {
      id: '0',
      type: 'function',
      function: {
        name: 'get_current_weather',
        arguments: '{"location":"Boston, MA","unit":"fahrenheit"}'
      }
    }
  ]
*/

API Reference:

Parallel tool calling​

The model may choose to call multiple tools. Here is an example using an extraction use-case:

import { z } from "zod";
import { zodToJsonSchema } from "zod-to-json-schema";

import { ChatAnthropicTools } from "@langchain/anthropic/experimental";
import { PromptTemplate } from "@langchain/core/prompts";
import { JsonOutputToolsParser } from "@langchain/core/output_parsers/openai_tools";

const EXTRACTION_TEMPLATE = `Extract and save the relevant entities mentioned in the following passage together with their properties.

Passage:
{input}
`;

const prompt = PromptTemplate.fromTemplate(EXTRACTION_TEMPLATE);

// Use Zod for easier schema declaration
const schema = z.object({
  name: z.string().describe("The name of a person"),
  height: z.number().describe("The person's height"),
  hairColor: z.optional(z.string()).describe("The person's hair color"),
});

const model = new ChatAnthropicTools({
  temperature: 0.1,
  model: "claude-3-sonnet-20240229",
}).bind({
  tools: [
    {
      type: "function",
      function: {
        name: "person",
        description: "Extracts the relevant people from the passage.",
        parameters: zodToJsonSchema(schema),
      },
    },
  ],
  // Can also set to "auto" to let the model choose a tool
  tool_choice: {
    type: "function",
    function: {
      name: "person",
    },
  },
});

// Use a JsonOutputToolsParser to get the parsed JSON response directly.
const chain = await prompt.pipe(model).pipe(new JsonOutputToolsParser());

const response = await chain.invoke({
  input:
    "Alex is 5 feet tall. Claudia is 1 foot taller than Alex and jumps higher than him. Claudia is a brunette and Alex is blonde.",
});

console.log(JSON.stringify(response, null, 2));

/*
  [
    {
      "type": "person",
      "args": {
        "name": "Alex",
        "height": 5,
        "hairColor": "blonde"
      }
    },
    {
      "type": "person",
      "args": {
        "name": "Claudia",
        "height": 6,
        "hairColor": "brunette"
      }
    }
  ]
*/

API Reference:

.withStructuredOutput({ ... })​

info

The .withStructuredOutput method is in beta. It is actively being worked on, so the API may change.

Using the .withStructuredOutput method, you can make the LLM return structured output, given only a Zod or JSON schema:

import { z } from "zod";
import { ChatAnthropicTools } from "@langchain/anthropic/experimental";
import { ChatPromptTemplate } from "@langchain/core/prompts";

const calculatorSchema = z.object({
  operation: z
    .enum(["add", "subtract", "multiply", "divide"])
    .describe("The type of operation to execute"),
  number1: z.number().describe("The first number to operate on."),
  number2: z.number().describe("The second number to operate on."),
});

const model = new ChatAnthropicTools({
  model: "claude-3-sonnet-20240229",
  temperature: 0.1,
});

// Pass the schema and tool name to the withStructuredOutput method
const modelWithTool = model.withStructuredOutput(calculatorSchema);

// You can also set force: false to allow the model scratchpad space.
// This may improve reasoning capabilities.
// const modelWithTool = model.withStructuredOutput(calculatorSchema, {
//   force: false,
// });

const prompt = ChatPromptTemplate.fromMessages([
  [
    "system",
    "You are a helpful assistant who always needs to use a calculator.",
  ],
  ["human", "{input}"],
]);

// Chain your prompt and model together
const chain = prompt.pipe(modelWithTool);

const response = await chain.invoke({
  input: "What is 2 + 2?",
});
console.log(response);
/*
  { operation: 'add', number1: 2, number2: 2 }
*/

API Reference:

Using JSON schema:​

import { ChatAnthropicTools } from "@langchain/anthropic/experimental";
import { ChatPromptTemplate } from "@langchain/core/prompts";

const calculatorJsonSchema = {
  type: "object",
  properties: {
    operation: {
      type: "string",
      enum: ["add", "subtract", "multiply", "divide"],
      description: "The type of operation to execute.",
    },
    number1: { type: "number", description: "The first number to operate on." },
    number2: {
      type: "number",
      description: "The second number to operate on.",
    },
  },
  required: ["operation", "number1", "number2"],
  description: "A simple calculator tool",
};

const model = new ChatAnthropicTools({
  model: "claude-3-sonnet-20240229",
  temperature: 0.1,
});

// Pass the schema and optionally, the tool name to the withStructuredOutput method
const modelWithTool = model.withStructuredOutput(calculatorJsonSchema, {
  name: "calculator",
});

const prompt = ChatPromptTemplate.fromMessages([
  [
    "system",
    "You are a helpful assistant who always needs to use a calculator.",
  ],
  ["human", "{input}"],
]);

// Chain your prompt and model together
const chain = prompt.pipe(modelWithTool);

const response = await chain.invoke({
  input: "What is 2 + 2?",
});
console.log(response);
/*
  { operation: 'add', number1: 2, number2: 2 }
*/

API Reference:

Help us out by providing feedback on this documentation page: