TaskForceAI SDK

Official SDK for the TaskForceAI multi-agent orchestration API.

Installation

npm install taskforceai-sdk

Runtime Requirements: Node.js 18 or newer (for built-in fetch and AbortController support).

Quick Start

import { TaskForceAI } from 'taskforceai-sdk';

const client = new TaskForceAI({
  apiKey: 'your-api-key-here',
});

// Submit a task (using default free model)
const taskId = await client.submitTask('Test prompt');

// Or submit with a specific model (Pro/Super plans)
const taskId2 = await client.submitTask('Complex analysis', {
  modelId: 'xai/grok-4.1',
});

// Wait for completion
const result = await client.waitForCompletion(taskId);
// handle result.result here

// Or stream status updates
const stream = client.streamTaskStatus(taskId);
for await (const status of stream) {
  // handle incremental status updates
}

Mock Mode

Build and test your integration without an API key using mock mode:

import { TaskForceAI } from 'taskforceai-sdk';

// No API key required in mock mode
const client = new TaskForceAI({ mockMode: true });

const result = await client.runTask('Test your integration');
console.log(result.result); // "This is a mock response. Configure your API key to get real results."

Mock mode simulates the full task lifecycle locally—no network requests are made. Tasks go through "processing" then "completed" states, making it easy to build UIs and test error handling before launch.

API Reference

TaskForceAI

Main SDK class for interacting with the TaskForceAI API.

Constructor

constructor(options: TaskForceAIOptions)

Options:

• apiKey (required unless mockMode is true): Your API key
• baseUrl (optional): API base URL (default: https://taskforceai.chat/api/developer)
• timeout (optional): Request timeout in milliseconds (default: 30000)
• responseHook (optional): Callback invoked with every raw fetch response
• mockMode (optional): Enable mock mode for development without an API key (default: false)

Methods

submitTask(prompt, options?)

Submit a task for multi-agent orchestration.

async submitTask(
  prompt: string,
  options?: TaskSubmissionOptions
): Promise<string>

Parameters:

• prompt: The user's input prompt
• options.modelId: Select a specific AI model (e.g., 'xai/grok-4.1', 'openai/gpt-5.2')
• options.silent: Suppress logging (default: false)
• options.mock: Use mock responses for testing (default: false)
• options.vercelAiKey: Supply your own Vercel AI Gateway API key (Pro/Super plans only)
• options.*: Any additional TaskForceAI orchestration flags are forwarded untouched

Returns: Task ID string

getTaskStatus(taskId)

Get the current status of a task.

async getTaskStatus(taskId: string): Promise<TaskStatus>

Returns: Object with taskId, status, result, and error fields

getTaskResult(taskId)

Get the final result of a completed task.

async getTaskResult(taskId: string): Promise<TaskResult>

Returns: Object with taskId and result fields

waitForCompletion(taskId, pollInterval?, maxAttempts?)

Wait for a task to complete by polling the status.

async waitForCompletion(
  taskId: string,
  pollInterval?: number,
  maxAttempts?: number,
  onStatus?: (status: TaskStatus) => void
): Promise<TaskResult>

Parameters:

• pollInterval: Milliseconds between polls (default: 2000)
• maxAttempts: Maximum polling attempts (default: 150)
• onStatus: Optional callback invoked each time a new status payload is fetched

runTask(prompt, options?, pollInterval?, maxAttempts?, onStatus?)

Submit a task and wait for completion in one call.

async runTask(
  prompt: string,
  options?: TaskSubmissionOptions,
  pollInterval?: number,
  maxAttempts?: number,
  onStatus?: (status: TaskStatus) => void
): Promise<TaskResult>

streamTaskStatus(taskId, pollInterval?, maxAttempts?, onStatus?)

Returns an AsyncIterable<TaskStatus> that yields each status payload. Useful for building progress UIs or logs. Call cancel() on the returned stream to stop polling locally without cancelling the backend computation.

runTaskStream(prompt, options?, pollInterval?, maxAttempts?, onStatus?)

Shortcut that submits a prompt and hands back a TaskStatusStream immediately so you can for await the updates without waiting for completion.

Error Handling

The SDK throws TaskForceAIError for API errors:

import { TaskForceAIError } from 'taskforceai-sdk';

try {
  const result = await client.runTask('Your prompt');
} catch (error) {
  if (error instanceof TaskForceAIError) {
    // handle API error, inspect error.statusCode and error.details
  }
}

Response Hooks & Telemetry

Pass responseHook to the constructor to observe every raw HTTP response and capture rate-limit headers or request IDs before the SDK parses them:

const client = new TaskForceAI({
  apiKey: 'key',
  responseHook: (response) => {
    console.log('x-ratelimit-remaining', response.headers.get('x-ratelimit-remaining'));
  },
});

Rate Limiting

The API includes built-in rate limiting. Check the response headers or use the SDK's error handling to manage rate limits.

Authentication

Get your API key from the TaskForceAI dashboard or contact support.

Examples

Basic Usage

import { TaskForceAI } from 'taskforceai-sdk';

const client = new TaskForceAI({ apiKey: 'your-key' });

const result = await client.runTask('What are the benefits of renewable energy?');
console.log(result.result);

Advanced Usage with Options

const taskId = await client.submitTask('Analyze this code for bugs', {
  modelId: 'xai/grok-4.1',
  silent: true,
  mock: false,
});

const status = await client.getTaskStatus(taskId);
console.log(`Status: ${status.status}`);

if (status.status === 'completed') {
  const result = await client.getTaskResult(taskId);
  console.log(result.result);
}

License

MIT