Vapi's create-squad skill helps an agent decide whether a voice use case needs several assistants at all and, if so, build a Vapi Squad with documented handoff tools between them. It defines each member's job, tools and boundaries, puts the entry member first, resolves real assistant IDs before wiring handoffs, and sets context and variable passing only where needed. By default it returns a plan or JSON payload and only makes live changes when you ask and a VAPI_API_KEY is set. It warns against inventing IDs, tools or credentials and against splitting each step of a conversation into its own assistant.
Build a Deepgram voice agent over one WebSocketThis skill from Deepgram shows an agent how to build a spoken assistant on Deepgram's agent API, where one WebSocket handles listening, the language model and speech output. It first helps decide between building on that API directly or using a framework such as Pipecat, LiveKit Agents, Retell or Vapi with Deepgram speech underneath. It then walks through the connection order, the Settings message, streaming audio, interruptions, changing settings during a call and function calls that your own code answers, including Twilio phone calls. It points to Deepgram's other skills for the full API reference and language SDKs.Voice agentsVoice agentsSpeech APIs
Create a Vapi voice assistantThis skill designs and builds a voice assistant on Vapi for phone or web calls. It covers the system prompt and first message, choosing a model, voice, and transcriber, multilingual setup, call-control tools, and hooks. It also explains errors from the create call. A Vapi API key is needed to save the assistant.Voice agentsVapiVoice AI
Give an OpenClaw agent a phone line with AmberAmber is a skill and runtime that gives an OpenClaw agent its own phone number through a bridge between Twilio, or a compatible voice provider, and OpenAI Realtime. It can answer and screen callers, place outbound calls only after confirmation, look up contacts, book on your calendar, take messages and log transcripts, and it exposes these as MCP tools plus a local call dashboard. The author describes it as a sensitive communications system and asks operators to decide how callers are told, how consent is gathered and how long data is kept before production use. The local CRM and transcript enrichment are off unless you enable them.Voice agentsVoice agentsOpenClaw
Scaffold a new Pipecat voice agent projectThis skill from the Pipecat team sets up a new Pipecat voice agent project by asking you a series of choices and then running the pipecat init command without its interactive prompts. It reads the current list of options from the CLI first, then asks for the project name, web or phone bot, client framework, transport, pipeline type and the speech, language model and voice services. It shows a summary before running the command and tells you to fill in your keys from the example environment file. If the Pipecat CLI is missing it tells you how to install it and stops.Voice agentsVoice agentsProject scaffolding
Blog Writing Guide (Sentry)Writes engineering blog posts in Sentry's voice—structure, examples, and editorial quality.Voice agents
Build voice agents with LiveKitThis skill helps you write spoken and text agents on the LiveKit Agents SDK and LiveKit Cloud. It focuses on keeping replies fast, keeping context small, and splitting one large agent into handoffs and tasks. It also covers fixes for agents that talk too much, repeat questions, or claim a booking that never saved.Voice agentsLiveKitVoice AI
