# HUMA Documentation

> HUMA is an async, stateful API for creating human-like AI agents that can participate in real-time interactions like games, chats, and voice calls.

## Documentation Structure

The documentation is organized into the following sections:

### Getting Started
- /docs - Documentation Overview
- /docs/quickstart - Quick Start Guide
- /docs/agent-lifecycle - Agent Lifecycle (Created → Connected → Active → Ended)

### Integration Guide
- /docs/integration-guide - Main Integration Guide (Mental Model)
- /docs/integration-wizard - Integration Wizard (Prompt Generator)

### Integration Steps (8 Phases)
1. /docs/integration-guide/01-personality - Design Personality (WHO the agent is)
2. /docs/integration-guide/02-rules - Design Rules (WHAT the agent does)
3. /docs/integration-guide/03-state - Design State (Context the agent receives)
4. /docs/integration-guide/04-events - State Changes & Events (Communication protocol)
5. /docs/integration-guide/05-tools - Design Tools (Actions agent can take)
6. /docs/integration-guide/06-router - Pick Router (Behavioral strategy selection)
7. /docs/integration-guide/07-voice-mode - Voice Mode (Enabling voice capabilities)
8. /docs/integration-guide/08-voice-lifecycle - Voice Lifecycle (Daily.co room management)

### Voice Implementation
- /docs/voice - Voice Implementation Guide (Daily.co, Deepgram, ElevenLabs integration)

### HUMA Models
- /docs/models/huma-1 - HUMA-1 (Server-side personalities)
- /docs/models/huma-0-1 - HUMA-0.1 (Client-defined agents with custom tools)

### API Reference
- /docs/api-reference - Complete API Reference

## Key Concepts

### Architecture
- Event-driven WebSocket communication
- Async tool execution (tools don't block agent processing)
- State is fully replaced on each context update

### Agent Types
- HUMA-1: Server-side personality storage, simpler setup
- HUMA-0.1: Client-defined personality/instructions/tools, more flexible

### Router Types
- llm-judge: General purpose, context-aware (default)
- conversational: Chat apps with timeliness scoring
- turn-taking: Multi-agent voice conversations
- voice-game: Voice gaming scenarios
- random: Testing, fast (no LLM call)

### Voice Integration
- Daily.co for real-time voice rooms
- Deepgram for speech-to-text
- ElevenLabs for text-to-speech
- Events: join-daily-room, leave-daily-room, voice-status, transcript, speak-status

## Quick Links

- Website: https://humalike.tech
- API Base URL: https://api.humalike.tech
- Documentation: https://humalike.tech/docs