init

Scaffold a new Pipecat project by collecting configuration from the user and running pc init in non-interactive mode.

Arguments

/init [--output <PATH>]

• --output (optional): Directory where the project will be created. Defaults to the current directory.

Prerequisites

Check if pc is installed by running pc --version. If not installed, tell the user to install it with uv tool install pipecat-ai-cli and stop.

Discover Available Options

Before asking the user any questions, run pc init --list-options to get the current valid values for all fields. The output is JSON:

{
  "bot_type": ["web", "telephony"],
  "transports": {
    "web": ["daily", "smallwebrtc"],
    "telephony": ["twilio", "telnyx", ...]
  },
  "stt": ["deepgram_stt", "openai_stt", ...],
  "llm": ["openai_llm", "anthropic_llm", ...],
  "tts": ["cartesia_tts", "elevenlabs_tts", ...],
  "realtime": ["openai_realtime", "gemini_live_realtime", ...],
  "video": ["heygen_video", "tavus_video", "simli_video"]
}

Use this data to populate the choices in every question below. Do NOT hardcode service lists — always use the values from --list-options.

Configuration Flow

Walk through the following questions to build the project configuration. After collecting all answers, show a summary and run the command.

Choosing the right interaction method:

• AskUserQuestion — Use for questions with a small, fixed set of options (bot type, pipeline mode, client framework, yes/no questions). This gives a clean clickable UI.
• Show list as text — Use for questions with many options (STT, LLM, TTS, realtime, video, transports). Display the full list of available options from --list-options formatted as a readable list, then let the user reply with their choice in chat.

Step 1: Project Name

Ask the user for a project name. This will be used as the directory name and project identifier.

Step 2: Bot Type

Ask the user to choose a bot type:

• Web/Mobile (web) - Browser or mobile app
• Telephony (telephony) - Phone calls