@agentai/agentai

Agent.ai JavaScript Library

A modern JavaScript library for interacting with the Agent.ai Actions API.

Installation

# Using npm
npm install @agentai/agentai

# Using yarn
yarn add @agentai/agentai

# Using pnpm
pnpm add @agentai/agentai

Example Usage

Before you can start using the library, you'll need to sign up for an account at Agent.ai and obtain a Bearer token from https://agent.ai/user/settings#credits.

JavaScript

import AgentAi from '@agentai/agentai';
// Using CommonJS
#const { AgentAiClient } = require('@agentai/agentai');

// Create a client instance with your bearer token
const client = new AgentAiClient('YOUR_BEARER_TOKEN_HERE');

// Example: Grab web text
async function fetchWebText() {
  try {
    const webTextResponse = await client.action(
      'grabWebText',
      { url: 'https://agent.ai' }
    );

    if (webTextResponse.status === 200) {
      console.log("Web Text Response Status:", webTextResponse.status);
      console.log("First 100 chars of Response:", webTextResponse.results.substring(0, 100) + "...");
    } else {
      console.error(`Error: Status Code: ${webTextResponse.status}, Message: ${webTextResponse.error}`);
    }
  } catch (error) {
    console.error('An error occurred:', error);
  }
}

// Example: Chat with an LLM
async function chatWithLlm() {
  try {
    const chatResponse = await client.chat(
      "What is an AI agent?", 
      { model: "gpt4o" }
    );

    if (chatResponse.status === 200) {
      console.log("Chat Response:", chatResponse.results);
    } else {
      console.error(`Error: ${chatResponse.error}`);
    }
  } catch (error) {
    console.error('An error occurred:', error);
  }
}

// Example: Get Google News
async function getGoogleNews() {
  try {
    const googleNewsResponse = await client.action(
      'getGoogleNews',
      {
        query: "AI advancements",
        date_range: "7d",
        location: "Boston"
      }
    );

    if (googleNewsResponse.status === 200) {
      console.log("Google News Location:", googleNewsResponse.metadata.search_information.location_used);
      console.log("Number of articles:", googleNewsResponse.results.length);
    } else {
      console.error(`Error: ${googleNewsResponse.error}`);
    }
  } catch (error) {
    console.error('An error occurred:', error);
  }
}

// Run the examples
fetchWebText();
chatWithLlm();
getGoogleNews();

API Reference

Creating a Client

// Method 1: Using API key only
const client = new AgentAiClient('YOUR_API_KEY');

// Method 2: Using API key with config object
const client = new AgentAiClient('YOUR_API_KEY', {
  timeout: 60000, // Optional: custom timeout in ms
  baseUrl: 'https://custom-url.com', // Optional: custom base URL
  headers: {
    'X-Custom-Header': 'value' // Optional: additional headers
  }
});

Methods

client.action(actionId, params)

Execute an AI action by its ID.

• actionId (string): The ID of the action to execute (e.g., 'grabWebText')
• params (object): Parameters for the action
• Returns: Promise

client.chat(prompt, options)

Use the invokeLlm action to generate text based on a prompt.

• prompt (string): The text prompt for the LLM
• options (object, optional):
  • model (string, default: 'gpt4o'): LLM model to use
  • Additional parameters for the invokeLlm action
• Returns: Promise

Response Structure

All methods return a Promise that resolves to an object with the following structure:

{
  status: 200,           // HTTP status code
  error: null,           // Error message (if any)
  results: {...},        // API response data (if successful)
  metadata: {...}        // Metadata from the API response (if available)
}

Error Handling

All methods return Promises that may reject with errors. It's recommended to use try/catch blocks when calling these methods:

try {
  const response = await client.action('grabWebText', { url: 'https://agent.ai' });
  if (response.status !== 200) {
    console.error(`API Error: ${response.error}`);
  }
} catch (error) {
  console.error('Network or client error:', error);
}

Contributing

Contributions are welcome! Please feel free to submit pull requests or open issues for feature requests or bug reports.

License

This project is licensed under the MIT License.