Jambonz starters

Skill jambonz/skills/skills/jambonz-starters

Ready-to-run jambonz starter applications. Use when the user wants to clone or scaffold a complete working voice app rather than assemble one from recipes — voice AI agents (OpenAI Realtime, Deepgram, ElevenLabs, Ultravox, Retell, Bedrock, AssemblyAI), IVR menus, dialers, recording, queuing, real-time translation, and more. Points at the @jambonz/sdk examples directory with one-line clone commands.From its SKILL.md

Install

npx -y skills add jambonz/skills --skill jambonz-starters

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

5.5 KB, as published. Nobody here has run it

jambonz Starter Applications

The @jambonz/sdk repository ships 18+ runnable starter apps in its examples/ directory. Each is a self-contained, copy-and-run project you can use as a base.

How to Choose a Starter

Just learning? → hello-world (webhook) — single say + hangup
Building a voice AI agent and have a vendor preference? → vendor-specific s2s starter (see "Voice AI Agents" table below)
Building a general voice agent? → agent (fully featured: OpenAI + Deepgram + tool calling)
Building an IVR / phone menu? → ivr-menu
Routing or transferring calls? → dial or two-stage-dialer
Need to record calls? → call-recording or listen-record
Support center / contact queue? → queue-with-hold
Something exotic? → see the full table below

Cloning a Starter

Use degit to grab a single example folder without the rest of the repo:

npx degit jambonz/node-sdk/examples/<starter-name> my-app
cd my-app
npm install

Then read the starter's app.ts (or ws-app.ts for WebSocket starters) and package.json to see the dependencies and entry point.

Voice AI Agents

Starter	Transport	What it shows
`agent`	WebSocket	Full voice agent: OpenAI LLM + Deepgram STT/TTS + Silero VAD + tool calling (`get_weather`)
`voice-agent`	WebSocket	Generic conversational AI voice agent powered by an LLM
`openai-realtime`	WebSocket	OpenAI Realtime API speech-to-speech integration
`deepgram-voice-agent`	WebSocket	Deepgram `deepgram_s2s` end-to-end
`elevenlabs-voice-agent`	WebSocket	ElevenLabs `elevenlabs_s2s` (uses `agent_id` auth — model & prompt live in the ElevenLabs dashboard)
`ultravox-voice-agent`	WebSocket	Ultravox `ultravox_s2s`
`retell-hosted`	Webhook	Retell-hosted agent via jambonz
`bedrock-agent`	WebSocket	AWS Bedrock-backed voice agent
`assemblyai-autogenerate-prompt`	WebSocket	AssemblyAI with dynamic prompt generation
`llm-streaming`	WebSocket	LLM token-streaming with `say` TTS

IVR, Routing, Call Control

Starter	Transport	What it shows
`hello-world`	Webhook + WS	The simplest jambonz application: answer, greet, hang up. Ships both a webhook (`app.ts`) and WebSocket (`ws-app.ts`) version side-by-side.
`ivr-menu`	Webhook	A classic interactive voice response menu that routes callers based on DTMF or speech input
`dial`	Webhook	Bridge an incoming call to a phone number, SIP endpoint, or registered user
`two-stage-dialer`	Webhook	Outbound dial that confirms the called party before bridging
`echo`	Webhook	Echo test — answers and plays back what the caller says

Recording, Queueing, Streaming

Starter	Transport	What it shows
`call-recording`	Webhook	Mid-call recording control: incoming call is bridged to an agent, recording can be started/stopped dynamically
`listen-record`	Webhook	Combine `listen` (audio streaming) with recording
`queue-with-hold`	Webhook	Support queue: callers placed in queue with hold music and position announcements. Agents connect separately to dequeue.
`realtime-translator`	WebSocket	Real-time bidirectional speech translation between two parties

After Cloning

Every starter expects a few application environment variables (set them in the jambonz portal under your application's settings). The starter's app.ts declares the expected variables at the top via envVars: { ... } — read that block, then create matching values in the portal.

For configuration patterns, load jambonz-recipes and read references/env-vars.md.