FunctionChain: OpenAI Function Calling Simplified in Node.js

Quickstart

This guide will walk you through the basic steps required to get the FunctionChain library up and running.

Installation

1. First of all, you need to clone the repository to your local machine:

git clone https://github.com/developersdigest/FunctionChain

2. Move to the directory where the repository has been cloned and install the necessary dependencies using npm:

cd FunctionChain
npm install

3. Rename the .env.example to .env at the root of your project. Obtain your OpenAI API Key from here, and add it to the .env file:

OPENAI_API_KEY=your_openai_api_key

4. Now you can run with the example in index.js by using:

node index.js
or
npm run dev

Setup

To setup FunctionChain, follow the steps below:

1. Create an index.js file in the root of your project.
2. Import the FunctionChain class from ai-function-chain and instantiate it.
3. Call the call method with a message. Optionally, you can specify a set of functions to execute.

import { FunctionChain, fetchCryptoPrice, openApp } from "ai-function-chain";

const functionChain = new FunctionChain({
  functions: [openApp, fetchCryptoPrice],
});

const res1 = await functionChain.call("Open the calculator on my computer");
const res2 = await functionChain.call("Get me the latest price of Bitcoin");
const res3 = await functionChain.call("Get me the latest price of Ethereum");

console.log(`${res1} \n${res2} \n${res3}`);

API Keys

To use FunctionChain, you must obtain and provide the appropriate API keys.

REQUIRED: for all functions:

OPENAI_API_KEY=your_openai_api_key

You need to obtain your OpenAI API Key here and add it to the .env file.

OPTIONAL: If you intend to use specific functions, you need to obtain the respective API keys:

For Alpha Vantage functions:

ALPHA_VANTAGE_API_KEY=your_alpha_vantage_api_key

Get your Alpha Vantage API key here

For Pinecone functions:

PINECONE_API_KEY=your_pinecone_api_key
PINECONE_ENVIRONMENT=your_pinecone_environment
PINECONE_INDEX=your_pinecone_index

Get your Pinecone API key here

For Visual Crossing API functions:

VISUAL_CROSSING_API_KEY=your_visual_crossing_api_key

Get your Visual Crossing API key here

For Huggingface API functions:

HUGGING_FACE_API_KEY=your_hugging_face_api_key

You can get your Huggingface API key here (A read only token)

Examples

Here are some examples of how you can use FunctionChain:

Example 1: Unix Example (No additional API Key Required)

import { FunctionChain, openApp } from "ai-function-chain";

const functionChain = new FunctionChain({functions: [openApp]});

const res = await functionChain.call("Open the calculator on my computer");

console.log(res);

Alternatively, you can run the following:

node examples/unix-example.js

Example 2: Crypto Prices (No API Key Required)

import { FunctionChain, fetchCryptoPrice } from "ai-function-chain";

const functionChain = new FunctionChain({functions: [fetchCryptoPrice]});

const res1 = await functionChain.call("Get me the latest price of Bitcoin");

const res2 = await functionChain.call("Get me the latest price of Ethereum");

console.log(`1. ${res1} \n2. ${res2}`);

Alternatively, you can run the following:

node examples/no-api-key-example.js

Example 3: Pinecone Example (Pinecone API Key Required)

You can get your Pinecone API Key here

import { FunctionChain, createPinecone, updatePinecone, askPinecone } from "ai-function-chain";

const functionChain = new FunctionChain({functions: [createPinecone, updatePinecone, askPinecone]});

const create = await functionChain.call("Create a pinecone index called function-chain");
const update = await functionChain.call("Add 'John Smith, his phone number is 123-456-7890 and email [email protected]' under the namespace my-contacts");
const ask = await functionChain.call("What is John Smiths number? to-do's in my my-contacts pinecone namespace?");

console.log(`1. ${create} \n2. ${update} \n3. ${ask}`);

Alternatively, you can run the following:

node examples/pinecone-example.js

Example 4: Alpha Vantage API (Free API Key Required)

You can get your Alpha Vantage API key here

import { FunctionChain, getAlphaVantageCompanyOverview } from "ai-function-chain";

const functionChain = new FunctionChain({functions: [getAlphaVantageCompanyOverview]});

const res1 = await functionChain.call("What is Apple's market capitalization");
const res2 = await functionChain.call("What is Microsoft's PE Ratio");
const res3 = await functionChain.call("What is Amazon's Revenue (TTM)");
const res4 = await functionChain.call("What is Alphabet's EBITDA");

console.log(`1. ${res1} \n2. ${res2} \n3. ${res3} \n4. ${res4}`);

Alternatively, you can run the following:

node examples/alpha-vantage-examples.js

Example 5: Huggingface Inference (Free API Key Required)

You can get your Huggingface API key here (A read only token)

import { FunctionChain, huggingFaceImageClassification } from "ai-function-chain";

const functionChain = new FunctionChain({functions: [huggingFaceImageClassification]}); 

const res = await functionChain.call("What is this image? https://www.shutterstock.com/image-photo/yellow-lovebird-sitting-alone-on-260nw-1894954309.jpg");
console.log(res);

Alternatively, you can run the following:

node examples/hugging-face-example.js

Example 6: Visual Crossing API (API Key Required)

You can get your Visual Crossing API key here

import { FunctionChain, getVisualCrossingWeatherForecast } from "ai-function-chain";

const functionChain = new FunctionChain({functions: [getVisualCrossingWeatherForecast]});

const res = await functionChain.call("What's the weather this week in Toronto");
console.log(res);

Alternatively, you can run the following:

node examples/weather-visual-crossing-example.js

Example 7: Using the Wikipedia API (No API Key required)

This example demonstrates how you can use the wikipedia function from the AI Function Chain to retrieve a summary of a Wikipedia page:

import { FunctionChain, wikipedia } from "ai-function-chain";

const functionChain = new FunctionChain({
  functions: [wikipedia],
  skipFinalAPICall: true,
});

const res = await functionChain.call("In one sentence, look up on wikipedia, what is Langchain?");
console.log(res);

You can also run this example directly from the command line:

node examples/wikipedia-example.js

Note: In this example, we're using the skipFinalAPICall: true option. This ensures that the result is obtained directly from the wikipedia function, without making a final API call to OpenAI.

Example 8: Using the Code Interpreter

This example demonstrates how you can use the codeInterpreter function from the AI Function Chain to execute code snippets in Node.js:

import { FunctionChain, codeInterpreter } from "ai-function-chain";

const functionChain = new FunctionChain({ functions: [codeInterpreter] });

let res1 = await functionChain.call("Using node.js execute a function that will give me a random number");
console.log(res1);

let res2 = await functionChain.call(`
  Execute this in Node.js:
  const fibonacci = (n) => {
    if (n <= 1) return n;
    return fibonacci(n - 1) + fibonacci(n - 2);
  };
  fibonacci(10);
`);
console.log(res2);

let res3 = await functionChain.call("Execute a twoSum function in node.js");
console.log(res3);

Function Descriptions

wikipedia

• Name: wikipedia
• Description: Search for "JavaScript" on Wikipedia.

codeInterpreter

• Name: codeInterpreter
• Description: Function can execute Node.js code directly code requested within a Node.js V8 runtime written in Javascript and returns the result. Doesn't support any new require or import statements yet

fetchCryptoPrice

• Name: fetchCryptoPrice
• Description: Fetches the price of a cryptocurrency from CoinGecko.

getAlphaVantageCompanyOverview

• Name: getAlphaVantageCompanyOverview
• Description: Fetches company information, financial ratios, and other key metrics for the specified equity using the Alpha Vantage API. This includes the following information: Symbol, Asset Type, Name, Description, CIK, Exchange, Currency, Country, Sector, Industry, Address, Fiscal Year End, Latest Quarter, Market Capitalization, EBITDA, PE Ratio, PEG Ratio, Book Value, Dividend Per Share, Dividend Yield, EPS, Revenue Per Share (TTM), Profit Margin, Operating Margin (TTM), Return on Assets (TTM), Return on Equity (TTM), Revenue (TTM), Gross Profit (TTM), Diluted EPS (TTM), Quarterly Earnings Growth (YoY), Quarterly Revenue Growth (YoY), Analyst Target Price, Trailing PE, Forward PE, Price to Sales Ratio (TTM), Price to Book Ratio, EV to Revenue, EV to EBITDA, Beta, 52-Week High, 52-Week Low, 50-Day Moving Average, 200-Day Moving Average, Shares Outstanding, Dividend Date, Ex-Dividend Date.

getAlphaVantageIntraday

• Name: getAlphaVantageIntraday
• Description: Fetches intraday data for a specified stock without extended hours.

askPinecone

• Name: askPinecone
• Description: This function queries/asks a question to a Pinecone index and returns the top answer. The Pinecone client, index name, and API keys are specified in the .env file or function parameters.

createPinecone

• Name: createPinecone
• Description: This function checks if a specified Pinecone index exists. If it does not exist, it will create a new one. Do not confuse this with updating an index.

updatePinecone

• Name: updatePinecone
• Description: This function updates a Pinecone index with vector embeddings generated from a given text with an optional namespace if passed. The Pinecone client, index name, and API keys are specified in the .env file or function parameters.

openApp

• Name: openApp
• Description: Opens a specified application on your computer

takeScreenshot

• Name: takeScreenshot
• Description: Captures a screenshot from the terminal

getVisualCrossingWeatherForecast

• Name: getVisualCrossingWeatherForecast
• Description: Fetches weather forecast for the specified location using the Visual Crossing API. This includes temperature, humidity, wind speed, and other important weather data.

Contribution

Contributions to the FunctionChain library are more than welcome! If you have any helpful functions you'd like to contribute, or if there's a library you'd like to see integrated with FunctionChain, please feel free to reach out or submit a pull request.

You can contribute in several ways:

• Submit a pull request: If you've written any functions that you'd like to share with the community, you can submit a pull request on the GitHub repository.
• Open an issue: If you've identified a bug or have a feature request, you can open an issue on the GitHub repository.
• Get in touch: If you have ideas or questions, feel free to reach out directly. Your feedback and ideas are invaluable in continuing to improve FunctionChain.

I am excited to see how you use FunctionChain and to hear your ideas for improvements!