Developer Documentation

Introduction

STACKOXA is a production-grade platform for building AI-native applications and multi-agent systems. The platform provides a unified SDK, CLI, and cloud runtime that handles agent lifecycle, memory management, model routing, and deployment.

This documentation covers the essential concepts and APIs you'll need to get started. For the full API reference, visit our developer portal (coming soon in public beta).

Quickstart

Get your first multi-agent system running in under 5 minutes:

# Install the CLI
npm install -g @stackoxa/cli

# Create a new project
stackoxa init my-first-agent
cd my-first-agent

# Start local development server
stackoxa dev

# Deploy to STACKOXA Cloud
stackoxa deploy

Installation

Requirements

• Node.js 20+ or Python 3.11+
• A STACKOXA account (free Starter tier available)
• API key from the STACKOXA dashboard

SDK Installation

# TypeScript / JavaScript
npm install @stackoxa/core

# Python
pip install stackoxa

Agents

Agents are the fundamental building blocks of STACKOXA applications. Each agent has a role, a set of tools, access to memory, and connection to one or more LLM providers.

import { Agent } from '@stackoxa/core'

const researcher = new Agent({
  name: 'researcher',
  role: 'Research and gather information on any topic',
  model: 'gpt-4o',
  tools: ['web_search', 'document_reader'],
  memory: { tier: 'hybrid' },
})

Memory Bus

The Memory Bus provides unified short-term and long-term memory for your agents. Short-term memory maintains conversation context within a session. Long-term memory uses vector embeddings for semantic retrieval across sessions.

• Ephemeral — In-memory context, cleared after session ends
• Persistent — Vector store backed, survives session restarts
• Hybrid — Combines both tiers with automatic promotion/demotion

Orchestration

Orchestration defines how multiple agents collaborate. STACKOXA supports sequential chains, parallel fan-out, conditional branching, and custom DAG workflows.

import { Stack, sequential } from '@stackoxa/core'

const pipeline = sequential([
  researcher,
  analyst,
  writer,
])

const result = await pipeline.run({
  input: 'Analyze Q1 market trends in AI infrastructure',
})

Deploy to Production

Deployment is a single command. STACKOXA handles containerization, health checks, auto-scaling, and rolling updates automatically.

stackoxa deploy --env production --region us-east-1

# Output:
# ✓ Built container image
# ✓ Pushed to registry
# ✓ Deployed 3 replicas
# ✓ Health checks passing
# → https://my-first-agent.stackoxa.app

Environment Variables

Configure your deployment with these environment variables:

• STACKOXA_API_KEY — Your platform API key
• STACKOXA_ENV — Deployment environment (development, staging, production)
• STACKOXA_LOG_LEVEL — Logging verbosity (debug, info, warn, error)
• OPENAI_API_KEY — Optional BYOK for OpenAI models
• ANTHROPIC_API_KEY — Optional BYOK for Anthropic models