@0latency/sdk

0Latency JavaScript/TypeScript SDK

A lightweight JavaScript/TypeScript client for the 0Latency agent memory API.

Installation

npm install @0latency/sdk

Quick Start

import { Memory } from '@0latency/sdk';

// Initialize the client
const memory = new Memory({ apiKey: 'your-api-key' });

// Add a memory
await memory.add('User prefers dark mode');

// Recall relevant memories
const results = await memory.recall('What are the user preferences?');
console.log(results);

Features

• TypeScript-first: Full type safety and IntelliSense support
• Zero dependencies: Uses native fetch API (Node.js 18+, all modern browsers)
• Simple API: Intuitive methods matching the Python SDK
• Error handling: Custom error classes for different failure modes
• Timeout support: Configurable request timeouts

API Reference

Constructor

const memory = new Memory({
  apiKey: 'your-api-key',          // Required
  baseUrl?: 'https://custom.api',  // Optional, defaults to api.0latency.ai
  timeout?: 30000                   // Optional, defaults to 30000ms
});

Methods

add(content, options?)

Store a new memory.

await memory.add('User loves coffee', {
  agentId: 'assistant-1',
  metadata: { category: 'preferences', confidence: 0.95 }
});

Parameters:

• content (string): The text content to remember
• options (optional):
  • agentId (string): Agent identifier for scoping
  • metadata (object): Key-value metadata to attach

Returns: API confirmation payload

---

recall(query, options?)

Retrieve relevant memories using semantic search.

const results = await memory.recall('What does the user like?', {
  agentId: 'assistant-1',
  limit: 5
});

Parameters:

• query (string): Natural-language search query
• options (optional):
  • agentId (string): Agent identifier for scoping
  • limit (number): Maximum results (default: 10)

Returns: Matching memories from the API

---

extract(conversation, options?)

Start async memory extraction from a conversation.

const result = await memory.extract([
  { role: 'user', content: 'I love hiking in the mountains' },
  { role: 'assistant', content: 'That sounds wonderful!' }
], { agentId: 'assistant-1' });

console.log('Job ID:', result.job_id);

Parameters:

• conversation (array): List of message objects with role and content
• options (optional):
  • agentId (string): Agent identifier

Returns: Object containing job_id for status polling

---

extractStatus(jobId)

Check the status of an extraction job.

const status = await memory.extractStatus('job-123');
console.log('Status:', status);

Parameters:

• jobId (string): The job identifier returned by extract()

Returns: Job status payload

---

health()

Check API health.

const health = await memory.health();
console.log('API status:', health);

Returns: Health status payload

Usage Examples

Simple Chatbot Memory

import { Memory } from '@0latency/sdk';

const memory = new Memory({ apiKey: process.env.ZEROLATENCY_API_KEY });

async function chatWithMemory(userMessage: string) {
  // Recall relevant context
  const context = await memory.recall(userMessage, {
    agentId: 'chatbot-1',
    limit: 3
  });

  // Generate response using context (pseudo-code)
  const response = await generateResponse(userMessage, context);

  // Store the conversation for future recall
  await memory.extract([
    { role: 'user', content: userMessage },
    { role: 'assistant', content: response }
  ], { agentId: 'chatbot-1' });

  return response;
}

Coding Agent with Project Memory

import { Memory } from '@0latency/sdk';

const memory = new Memory({ apiKey: process.env.ZEROLATENCY_API_KEY });

async function codingAgent(projectId: string, userRequest: string) {
  // Recall project-specific knowledge
  const projectKnowledge = await memory.recall(
    `${userRequest} in the context of this project`,
    { agentId: `project-${projectId}`, limit: 5 }
  );

  // Store important facts about the project
  await memory.add('User prefers functional programming style', {
    agentId: `project-${projectId}`,
    metadata: { 
      category: 'coding-style',
      project: projectId,
      timestamp: Date.now()
    }
  });

  return projectKnowledge;
}

Background Extraction

import { Memory } from '@0latency/sdk';

const memory = new Memory({ apiKey: process.env.ZEROLATENCY_API_KEY });

async function extractAndPoll(conversation: any[]) {
  // Start extraction job
  const { job_id } = await memory.extract(conversation, {
    agentId: 'assistant-1'
  });

  // Poll for completion
  let status;
  do {
    await new Promise(resolve => setTimeout(resolve, 1000)); // Wait 1 second
    status = await memory.extractStatus(job_id);
  } while (status.state === 'processing');

  console.log('Extraction complete:', status);
}

Error Handling

The SDK provides custom error classes for different failure modes:

import { 
  Memory, 
  AuthenticationError, 
  RateLimitError, 
  ZeroLatencyError 
} from '@0latency/sdk';

const memory = new Memory({ apiKey: 'your-api-key' });

try {
  await memory.add('Some content');
} catch (error) {
  if (error instanceof AuthenticationError) {
    console.error('Invalid API key');
  } else if (error instanceof RateLimitError) {
    console.error('Rate limit exceeded, retry later');
  } else if (error instanceof ZeroLatencyError) {
    console.error('API error:', error.message, error.statusCode);
  } else {
    console.error('Unexpected error:', error);
  }
}

TypeScript Support

Full TypeScript definitions are included. Import types as needed:

import type { 
  MemoryConfig, 
  AddMemoryOptions,
  RecallOptions,
  ConversationMessage 
} from '@0latency/sdk';

Requirements

• Node.js: 18.0.0 or higher (for native fetch support)
• Browsers: All modern browsers with fetch support

For older Node.js versions, you can polyfill fetch using node-fetch or undici.

License

MIT License - see LICENSE file for details.

Support

• Email: justin@0latency.ai
• Documentation: docs.0latency.ai
• Issues: GitHub Issues

Contributing

Contributions welcome! Please open an issue or PR on GitHub.