Prompter

Personal Voice Assistant CLI with teleprompter-style display. Speak questions, get AI-powered answers displayed in large, easy-to-read text.

Features

• Push-to-talk voice input - Press Space to record, Space again to stop
• Speech-to-text via OpenAI Whisper
• AI responses via Claude Opus 4.5 (through OpenRouter)
• Streaming teleprompter display - Answers displayed word-by-word
• Learning mode - Automatically extracts and remembers facts about you
• Personal memory - Stores facts, conversations, and knowledge base
• User profiles - Customize response style and personal context

Installation

npm install -g voice-prompter

Quick Setup

1. Install audio recording tool

macOS:

brew install sox

Ubuntu/Debian:

sudo apt-get install sox

Windows:

choco install sox

2. Set API keys

# OpenRouter (for Claude Opus 4.5)
export OPENROUTER_API_KEY=sk-or-...

# OpenAI (for Whisper transcription)
export OPENAI_API_KEY=sk-...

3. Verify setup

prompter setup

4. Start using

prompter

Usage

Interactive Mode

prompter

Keyboard Controls: | Key | Action | |-----|--------| | Space | Push to talk (press to start, press again to stop) | | L | Toggle learning mode | | C | Toggle continuous listening | | P | Edit profile | | K | Add knowledge | | Q | Quit |

CLI Commands

prompter              # Start interactive mode
prompter setup        # Check dependencies and setup status
prompter profile      # View your profile
prompter learn FILE   # Add a document to knowledge base
prompter config       # View configuration
prompter clear-memory # Clear all stored memory

How It Works

┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│   Speak     │────▶│   Whisper   │────▶│   Claude    │
│  Question   │     │   (OpenAI)  │     │  Opus 4.5   │
└─────────────┘     └─────────────┘     └─────────────┘
                                               │
                                               ▼
┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│   Learn     │◀────│   Display   │◀────│   Answer    │
│   Facts     │     │ Teleprompter│     │  Streamed   │
└─────────────┘     └─────────────┘     └─────────────┘

1. Voice Capture - Records audio via SoX
2. Transcription - Sends audio to OpenAI Whisper API
3. Context Building - Combines your profile, facts, recent conversations, and knowledge
4. AI Response - Sends question + context to Claude Opus 4.5 via OpenRouter
5. Display - Streams response word-by-word in teleprompter style
6. Learning - Extracts and stores facts from conversation

Data Storage

All data is stored locally in ~/.prompter/:

~/.prompter/
├── config.json           # App settings
├── profiles/
│   └── default.json      # User profile
└── memory/
    ├── facts.json        # Learned facts about you
    ├── conversations.json # Conversation history
    └── knowledge/        # Ingested documents

Getting API Keys

OpenRouter (for Claude)

1. Go to openrouter.ai
2. Sign up and add credits
3. Create an API key starting with sk-or-

OpenAI (for Whisper)

1. Go to platform.openai.com
2. Sign up and add credits
3. Create an API key starting with sk-

Requirements

• Node.js 18+
• SoX (audio recording)
• Microphone access
• API keys for OpenRouter and OpenAI

Troubleshooting

"No audio recording tool found"

Install SoX using the commands in the setup section above.

"Could not understand audio"

• Speak clearly and at normal volume
• Check your microphone is working
• Ensure you're in a quiet environment

API errors

• Verify API keys are set: prompter setup
• Check you have credits on both OpenRouter and OpenAI
• Verify network connectivity

Permission errors on macOS

Grant terminal microphone permission in System Preferences > Security & Privacy > Privacy > Microphone.

License

MIT