Inworld Runtime Templates

Templates for building AI applications with the Inworld Runtime SDK. From simple LLM calls to complete voice agents, these templates demonstrate best practices and advanced patterns for conversational AI, RAG, voice processing, safety, and more.

Prerequisites

• npm or Yarn
• Node.js v20 or higher: Download here
• Inworld API key (required): Sign up here or see quickstart guide

Get Started

Step 1: Clone the Repository

git clone https://github.com/inworld-ai/inworld-runtime-templates-node.git
cd inworld-runtime-nodejs-templates

Step 2: Install Dependencies

npm install
# or
yarn install

Step 3: Configure Environment Variables

Create a .env file in the root directory:

INWORLD_API_KEY=your_api_key_here

Get your API key from the Inworld Portal.

Alternatively, export the key as an environment variable:

export INWORLD_API_KEY="your_api_key_here"

Optional Environment Variables

Variable	Default	Description
INWORLD_ADDON_POOL_SIZE	64	Thread pool size for native addon. Increase for high-concurrency workloads.

# For running multiple concurrent graphs or load testing
export INWORLD_ADDON_POOL_SIZE=128

Step 4: Run the Templates

Run any template using the scripts defined in package.json:

# Basic LLM chat
npm run node-llm-chat "Tell me about AI"

# Speech to text
npm run node-stt path/to/audio.wav

# Intent matching
npm run node-intent "What house are you in?"

Repo Structure

inworld-runtime-nodejs-templates/
├── src/                    # All template source files
│   ├── llm/                # LLM operations, routing, and tools
│   │   ├── routing/        # Intelligent routing and conditional logic
│   │   └── tools/          # Function calling and MCP integration
│   ├── voice/              # Voice processing (STT, TTS)
│   ├── advanced/           # Advanced examples and patterns
│   ├── retrieval/          # RAG, search, intent matching, and memory
│   ├── safety/             # Content moderation and filtering
│   ├── streaming/          # Real-time streaming for text and audio
│   ├── text_processing/    # Text transformation and template rendering
│   ├── observability/      # Telemetry, metrics, and monitoring
│   ├── primitives/         # Low-level SDK primitives for direct usage
│   ├── utilities/          # Utility nodes and helpers
│   └── shared/             # Shared utilities and helpers
├── package.json            # Dependencies and scripts
├── README.md               # Documentation
└── LICENSE                 # MIT License

Templates Overview

LLM Templates (src/llm/)

• Basic Chat & Completion: Simple LLM interactions
• Routing: Intelligent routing and conditional logic
• Tools: Function calling and MCP (Model Context Protocol) integration
• Component Registry: Advanced component management patterns

Voice Templates (src/voice/)

• Speech-to-Text (STT): Convert audio to text
• Text-to-Speech (TTS): Generate audio from text with streaming support

Retrieval Templates (src/retrieval/)

• RAG: Knowledge retrieval and question answering
• Intent Matching: Classify user intents and route accordingly
• Long-term Memory: Persistent conversation memory

Safety Templates (src/safety/)

• Content Moderation: Keyword matching and text classification
• Safety Pipelines: Complete content filtering workflows

Streaming Templates (src/streaming/)

• LLM Streaming: Real-time text generation with abort controls
• Audio Streaming: Process audio streams in real-time

Primitives (src/primitives/)

Low-level SDK components for direct usage:

• Embedder: Generate text embeddings
• NER: Named entity recognition
• VAD: Voice activity detection
• Jinja Templates: Template rendering

Learning Path

New to Inworld Runtime?

1. Start with src/llm/llm_chat.ts - Basic LLM chat
2. Try src/voice/speech_to_text.ts - Voice processing
3. Explore src/streaming/ - Real-time patterns

Building a Chatbot?

1. src/llm/llm_chat.ts - Basic LLM chat
2. src/llm/tools/mcp_call_tool.ts - Add function calling
3. src/llm/routing/llm_output_routing.ts - Add routing logic
4. src/safety/safety_pipeline.ts - Add content moderation

Building a Voice Agent?

1. src/voice/speech_to_text.ts - STT
2. src/voice/text_to_speech.ts - TTS

Adding RAG?

1. src/retrieval/knowledge_retrieval.ts - Basic RAG
2. src/retrieval/intent_matching.ts - Intent matching
3. src/retrieval/knowledge_routing.ts - Route based on context

Troubleshooting

Missing API Key

export INWORLD_API_KEY="your_key_here"
# Or create a .env file

Module Not Found

npm install
# or
yarn install

Audio Issues

Grant microphone permissions in your browser for voice examples.

Bug Reports: GitHub Issues

General Questions: For general inquiries and support, please email us at [email protected]

Contributing

We welcome contributions! Found a bug or want to add a template? Open an issue or PR, and we will review it.

For detailed guidelines, please see CONTRIBUTING.md.

License

This project is licensed under the MIT License - see the LICENSE file for details.