@embedapi/core

EmbedAPIClient

🔥 ONE API KEY TO RULE THEM ALL! Access ANY AI model instantly through our game-changing unified API. Build AI apps in minutes, not months! The ultimate all-in-one AI agent solution you've been waiting for! 🚀

EmbedAPI is a powerful Node.js client library that gives you instant access to ANY AI model through a single, unified API. Stop juggling multiple API keys and implementations - with EmbedAPI you can build sophisticated AI applications in minutes, not months.

Installation

npm install @embedapi/client

Initialization

const EmbedAPIClient = require('@embedapi/client');

// Regular API client
const client = new EmbedAPIClient('your-api-key');

// Agent mode client
const agentClient = new EmbedAPIClient('your-agent-id', { isAgent: true });

// Debug mode client
const debugClient = new EmbedAPIClient('your-api-key', { debug: true });

// Agent and debug mode client
const debugAgentClient = new EmbedAPIClient('your-agent-id', { isAgent: true, debug: true });

Constructor Parameters

• apiKey (string): Your API key for regular mode, or agent ID for agent mode
• options (object, optional): Configuration options
  • isAgent (boolean, optional): Set to true to use agent mode. Defaults to false
  • debug (boolean, optional): Set to true to enable debug logging. Defaults to false

Methods

1. generate({ service, model, messages, ...options })

Generates text using the specified AI service and model.

Parameters

• service (string): The name of the AI service (e.g., 'openai')
• model (string): The model to use (e.g., 'gpt-4o')
• messages (array): An array of message objects containing conversation history
• maxTokens (number, optional): Maximum number of tokens to generate
• temperature (number, optional): Sampling temperature
• topP (number, optional): Top-p sampling parameter
• frequencyPenalty (number, optional): Frequency penalty parameter
• presencePenalty (number, optional): Presence penalty parameter
• stopSequences (array, optional): Stop sequences for controlling response generation
• tools (array, optional): Array of function definitions for tool use
• toolChoice (string|object, optional): Tool selection preferences
• enabledTools (array, optional): List of enabled tool names
• userId (string, optional): Optional user identifier for agent mode

Usage Example

// Regular mode
const response = await client.generate({
    service: 'openai',
    model: 'gpt-4o',
    messages: [{ role: 'user', content: 'Hello' }]
});

// Agent mode
const agentResponse = await agentClient.generate({
    service: 'openai',
    model: 'gpt-4o',
    messages: [{ role: 'user', content: 'Hello' }]
});

2. stream({ service, model, messages, ...options })

Streams text generation using the specified AI service and model.

Parameters

Same as generate(), plus:

• streamOptions (object, optional): Stream-specific configuration options

Response Format

The stream emits Server-Sent Events (SSE) with two types of messages:

• Content Chunks:

{
    "content": "Generated text chunk",
    "role": "assistant"
}

• Final Statistics:

{
    "type": "done",
    "tokenUsage": 17,
    "cost": 0.000612
}

Usage Example

// Regular mode
const streamResponse = await client.stream({
    service: 'openai',
    model: 'gpt-4o',
    messages: [{ role: 'user', content: 'Hello' }]
});

// Agent mode
const agentStreamResponse = await agentClient.stream({
    service: 'openai',
    model: 'gpt-4o',
    messages: [{ role: 'user', content: 'Hello' }]
});

// Process the stream
const reader = streamResponse.body.getReader();
const decoder = new TextDecoder();

while (true) {
    const { done, value } = await reader.read();
    if (done) break;
    
    const chunk = decoder.decode(value);
    const lines = chunk.split('\n');
    
    for (const line of lines) {
        if (line.startsWith('data: ')) {
            const data = JSON.parse(line.slice(6));
            if (data.type === 'done') {
                console.log('Stream stats:', {
                    tokenUsage: data.tokenUsage,
                    cost: data.cost
                });
            } else {
                console.log('Content:', data.content);
            }
        }
    }
}

3. listModels()

Lists all available models.

const models = await client.listModels();

4. testAPIConnection()

Tests the connection to the API.

const isConnected = await client.testAPIConnection();

Error Handling

All methods throw errors if the API request fails:

try {
    const response = await client.generate({
        service: 'openai',
        model: 'gpt-4o',
        messages: [{ role: 'user', content: 'Hello' }]
    });
} catch (error) {
    console.error('Error:', error.message);
}

Authentication

The client supports two authentication modes:

• Regular Mode (default)

  • Uses API key in request headers
  • Initialize with: new EmbedAPIClient('your-api-key')

• Agent Mode

  • Uses agent ID in request body
  • Initialize with: new EmbedAPIClient('your-agent-id', { isAgent: true })
  • Optional userId parameter available for request tracking

License

MIT